Functional Design of a Hybrid Leg-Wheel-Track Ground Mobile Robot - MDPI
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
machines Article Functional Design of a Hybrid Leg-Wheel-Track Ground Mobile Robot Luca Bruzzone * , Mario Baggetta, Shahab E. Nodehi , Pietro Bilancia and Pietro Fanghella Department of Mechanical, Energy, Management and Transportation Engineering, University of Genoa, 16145 Genoa, Italy; mario.baggetta@unige.it (M.B.); shahabedin.nodehi@edu.unige.it (S.E.N.); pietro.bilancia@edu.unige.it (P.B.); pietro.fanghella@unige.it (P.F.) * Correspondence: luca.bruzzone@unige.it; Tel.: +39-010-335-2967 Abstract: This paper presents the conceptual and functional design of a novel hybrid leg-wheel- track ground mobile robot for surveillance and inspection, named WheTLHLoc (Wheel-Track-Leg Hybrid Locomotion). The aim of the work is the development of a general-purpose platform capable of combining tracked locomotion on irregular and yielding terrains, wheeled locomotion with high energy efficiency on flat and compact grounds, and stair climbing/descent ability. The architecture of the hybrid locomotion system is firstly outlined, then the validation of its stair climbing maneuver capabilities by means of multibody simulation is presented. The embodiment design and the internal mechanical layout are then discussed. Keywords: ground mobile robot; hybrid locomotion; step climbing 1. Introduction In the last decade, the world market of service robotics has increased remarkably, overcoming the market of industrial robotics [1]. In particular, ground mobile robots are the most widespread category of service robots, with important application fields Citation: Bruzzone, L.; Baggetta, M.; such as precision agriculture [2], planetary exploration [3], homeland security [4], surveil- Nodehi, S.E.; Bilancia, P.; Fanghella, P. lance [5], reconnaissance, and intervention in case of terroristic attacks and in the presence Functional Design of a Hybrid of radioactive or chemical contamination [6]. Leg-Wheel-Track Ground Mobile Independently of the robot payload, which depends on the specific application, a Robot. Machines 2021, 9, 10. https:// doi.org/10.3390/machines9010010 great research effort has been spent to develop innovative locomotion systems for ground mobile robots [7–9]. While locomotion on flat and even terrain can be simply and effectively Received: 10 November 2020 performed by wheels (W) with high speed and energy efficiency [10], in case of unstructured Accepted: 7 January 2021 environments, the use of tracks (T) and legs (L) is also a valuable option. Published: 12 January 2021 Tracked robots can move on uneven and yielding terrains, since their contact surface with the ground is much larger than the one of wheeled robots [11–13]. On the other hand, Publisher’s Note: MDPI stays neu- they usually move slower and with lower energy efficiency. In addition, they are more tral with regard to jurisdictional clai- subject to vibrations with respect to wheeled robots. ms in published maps and institutio- Legged locomotion, which is, differently from wheeled and tracked locomotion, nal affiliations. biologically inspired, is the most effective solution in case of irregular terrains and obstacles, but it is generally slower and less energetically efficient on flat grounds; the main hindrance to the diffusion of legged robots is their cost and complexity, especially for the ones with dynamic gait [14,15]; the higher complexity of the dynamic gait is related not only Copyright: © 2021 by the authors. Li- to control, but also to the mechanical architecture, characterized by a high number of censee MDPI, Basel, Switzerland. This article is an open access article actuators. To overcome this issue, at least for small mobile robots, with limited structural distributed under the terms and con- stresses due to the inertial effects, the complexity of the leg architecture can be simplified: ditions of the Creative Commons At- Examples are robots with rotating and compliant legs [16,17], or hybrid leg–wheel robots tribution (CC BY) license (https:// with stepping triple wheels [18,19]. creativecommons.org/licenses/by/ Hybrid locomotion systems combine the benefits of different locomotion systems. 4.0/). One of the most advanced and impressive hybrid locomotion robots is the legged-wheeled Machines 2021, 9, 10. https://doi.org/10.3390/machines9010010 https://www.mdpi.com/journal/machines
Machines 2021, 9, 10 2 of 11 Handle by Boston Dynamics [20], which is a biped with two wheels at the end of the legs; a similar concept has been adopted in the design of the Centauro quadruped, by IIT [21]. At a smaller scale, there are many examples of hybrid robots which combine legs, wheels, and tracks in a simplified manner, with lower cost and consequently a wider range of potential applications: the four possible combinations of L, W and T (LW, LT, WT, LWT) are compared in [8] in terms of maximum speed, obstacle crossing capability, step/stair climb- ing capability, walking capability of soft terrains and on uneven terrains, energy efficiency, mechanical, and control complexity. Considering locomotion solutions involving tracks (T, WT, LT, LWT), there are many possible layouts. Robots can have non-articulated or articulated tracks with passive relative mobility [22]; if the relative mobility of the articulated tracks is actuated to improve obstacle crossing capability, they can be considered as LT hybrid robots. Another option is to link tracked modules in series to compose snake-like climbing robots [23]. Focusing on LT hybrid robots, there are different ways to combine legs and tracks. Besides the most obvious approach of using them in parallel [24–26], peripheral tracks can be placed on the leg links [27]; this approach is also used for the Kylin robot, with LWT hybrid locomotion [28]. When wheels are present and it is required to use them to increase speed and range on even and compact grounds, they can be activated using many different systems. For ex- ample, the contact of the wheels with the terrain can be enabled or disabled by the position of the legs, as in the LWT Kylin robot. Another possibility is to extract the wheels from the robot body to suspend the robot on wheels. This solution has the benefit of enhancing the vehicle maneuverability, which is a drawback of purely tracked robots; an example is the hybrid WT robot presented in [29]. In a previous study [30], a WT robot with two variable-shape tracks and four wheels is proposed; the contact between wheels and terrain can be enabled or disabled by varying the shape of the tracks. The hybrid LWT Azimuth robot [31] is equipped with articulated legs with peripheral tracks and wheels; depending on the leg position, wheels, legs, or tracks can be active. The hybrid WT robot HELIOS-VI [32] has two tracks and two passive wheels placed at the end of a rotating arm, which regulates the contact of the wheels with the terrain to reduce the contact areas of the tracks, to improve maneuverability or to perform stair climbing and descent. In other cases, the tracks can be transformed into wheels changing their outer shape by means of tensioning systems, exploiting the tracks elasticity [33,34]; the transformation of tracks into wheels can also be achieved by adopting complex articulated supports for the tracks [35]. This research is focused on the development of a hybrid LWT ground mobile robot for surveillance and inspection, named WheTLHLoc (Wheel-Track-Leg Hybrid Locomotion). Its locomotion system shows similarities with previously discussed robots, although the proposed solution is new and has unique features with respect to other solutions presented in scientific literature; similarities and differences will be highlighted in Section 2. The con- sidered size is small (around 450 × 350 × 130 mm), since only cameras and environmental sensors are to be carried and good maneuverability in restricted spaces is considered a key feature of an inspection robot. Despite its limited dimensions, the combination of all the three locomotion systems allows to perform wheeled locomotion with high speed, energy efficiency, and range on flat and compact grounds, tracked locomotion on yielding and soft terrains, and mixed use of rotating legs, wheels, and tracks to overcome steps, stairs, and other obstacles. The remaining of the paper is organized as follows: Section 2 discusses the conceptual design and the main functional features of the WheTLHLoc robot; Section 3 illustrates the maneuvers for stair climbing and descent, and the feasibility of the most critical one (climbing) is verified by multibody simulation in Section 4; Section 5 presents the
maneuvers for stair climbing and descent, and the feasibility of the most critical one (climbing) is verified by multibody simulation in Section 4; Section 5 presents the embod- iment design and the internal layout of the robot, and finally Section 6 outlines conclu- sions and future work. Machines 2021, 9, 10 3 of 11 2. Conceptual Design of the Hybrid Robot WheTLHLoc The starting point of the conceptual design of the proposed hybrid locomotion robot is the goal of combining tracks for soft and yielding terrains and wheeled locomotion on embodiment design and the internal layout of the robot, and finally Section 6 outlines flat and compact terrains. This combination can be obtained in different ways, as dis- conclusions and future work. cussed in the previous section. The proposed locomotion system (Figure 1) is in some ways 2. similar toDesign Conceptual the oneofofthe HELIOS-VI [32], but Hybrid Robot enhances its capabilities thanks to the fol- WheTLHLoc lowing design choices: The starting point of the conceptual design of the proposed hybrid locomotion robot • theInstead is goal ofof a single short combining tracks rotating for softarmandplaced yielding on terrains the frontandof the vehicle, wheeled it makes use locomotion on of two independently actuated rotating legs (Figure 1, L); flat and compact terrains. This combination can be obtained in different ways, as discussed • the in The wheels section. previous at the legTheendsproposed (Figure 1,locomotion W) are not passive, but independently system (Figure 1) is in some actuated ways by to similar gearmotors the one of (Figure HELIOS-VI 1, WM);[32], but enhances its capabilities thanks to the following • designThe common revolute axis of the two legs (Figure 1, A), is centered with respect to choices: • Instead ofbody, the robot a singleandshort the legs can perform rotating arm placed complete on the and frontcontinuous rotations of the vehicle, around it makes use it; two independently actuated rotating legs (Figure 1, L); of •• Twowheels The omni wheels (Figure at the leg ends 1, OW) are (Figure placed 1, W) on passive, are not the rear ofbutthe robot, so thatactuated independently the con- tactgearmotors by of the tracks (Figure (Figure 1, 1, T) with the terrain is avoided when the robot is supported WM); • on the The two wheels common andaxis revolute on one of two of the the omni wheels;1, A), is centered with respect to the legs (Figure • robot body, and the legs can perform complete and symmetric The main robot body (Figure 1, MB) is externally continuouswith respect rotations to theit;xy around • plane. Two omni wheels (Figure 1, OW) are placed on the rear of the robot, so that the contact of The themain tracks (Figure camera is 1, T) with placed on the the terrain front ofisthe avoided robot when (Figurethe1,robot is supported FC), but a secondaryon the two wheels and on one of the omni wheels; camera is placed on the rear (Figure 1, RC) since during step climbing (Section 3) and in •otherThe main robot scenarios body (Figure it is necessary to move 1, MB) is externally backwards; symmetric additional with respect environmental to the sensors can xy plane. be placed inside the robot body, on the basis of the required tasks. Figure Figure 1. Conceptualdesign 1. Conceptual designofofthe thehybrid hybridleg-wheel-track leg-wheel-track robot robot Wheel-Track-Leg Wheel-Track-Leg Hybrid Hybrid Locomotion Locomotion (WheTLHLoc): (WheTLHLoc): (a) (a) Front Front view; view; (b) rear (b) rear viewview (W: Wheel; (W: Wheel; L:T:Leg; L: Leg; T: Track; Track; MB:body; MB: Main MainWM: body; WM:gearmotor; Wheel Wheel gearmotor; OW: OmniOW: Omni wheel; FC:wheel; Front camera; FC: FrontRC: Rear camera; camera; RC: RearA: Leg axis). camera; A: Leg axis). The main features camera is ofplaced on the this design arefront of the robot (Figure 1, FC), but a secondary the following: camera is placed on the rear (Figure 1, RC) since during step climbing (Section 3) and in other scenarios it is necessary to move backwards; additional environmental sensors can be placed inside the robot body, on the basis of the required tasks. The main features of this design are the following: i. It is possible to alternate wheeled locomotion on flat and compact terrains (Figure 2a) and tracked locomotion on soft and yielding terrains (Figure 2b), both with differential steering; the switch between the locomotion modes is commanded by the rotation of the legs; ii. The legs, longer than the arm of HELIOS-VI, can be used to overcome obstacles, even performing differential steering, if required in case of asymmetrical obstacles;
i. It is possible to alternate wheeled locomotion on flat and compact terrains (Figure 2a) and tracked locomotion on soft and yielding terrains (Figure 2b), both with differen- Machines 2021, 9, 10 tial steering; the switch between the locomotion modes is commanded by the rotation 4 of 11 of the legs; ii. The legs, longer than the arm of HELIOS-VI, can be used to overcome obstacles, even performing differential steering, if required in case of asymmetrical obstacles; iii. iii. When Whenthe therobot robot isis in in the the wheeled wheeled locomotion locomotion (WL) (WL) position, position, thethe field field ofof view view ofof the the front camera is optimally exploited front camera is optimally exploited (Figure 2a); (Figure 2a); iv. iv. Since Sincethetherobot robotmainmainbody body is symmetric is symmetric withwith respect respect to theto xy xy plane, theplane, it is oper- it is fully fully operative even after an overturn; it is very unlikely that after a ative even after an overturn; it is very unlikely that after a fall the robot remains on fall the robot remains on one one flank, flank, andand eveneven in this in this case case a rotation a rotation ofofthe theleg leginincontact contactwith with thethe terrain terrain can can putagain put again thethe robot robot on on the the tracks; tracks; v. v. The rotation of the legs The rotation of the legs and the and the action action ofof the the wheels wheels can can bebe used used inin case case ofof irregular irregular terrains and obstacles, i.e., when the use of only tracks is not terrains and obstacles, i.e., when the use of only tracks is not sufficient to advance; sufficient to advance; in in this this case, case, thethetwotwo legscan legs canbebemoved movedindependently independentlyin in presence presence of of asymmetric asymmetric groundirregularities ground irregularities (Figure (Figure 2c); 2c); vi. vi. In particular, the combined motionofoflegs, In particular, the combined motion wheels, legs, wheels, and tracks and cancan tracks be used to overcome be used to over- steps and stairs, according to the sequences discussed come steps and stairs, according to the sequences discussed in Section 3. in Section 3. Figure 2. (a) (a)Wheeled Wheeledlocomotion locomotionon onflat flatand andcompact surfaces; compact (b)(b) surfaces; tracked locomotion tracked on yielding locomotion terrains; on yielding (c) hybrid terrains; leg- (c) hybrid wheel-track locomotion leg-wheel-track to overcome locomotion obstacles. to overcome obstacles. previously mentioned, the proposed locomotion system presents similarities with As previously the one the one of of HELIOS-VI, HELIOS-VI,though thoughititshows showsthe thefollowing followingdifferences: differences: (i)(i) Independently Independently ac- actu- tuated and longer legs, which can perform complete and ated and longer legs, which can perform complete and continuous rotations in continuous rotations in case of obstacle overcoming obstacle overcoming and and stair climbing; (ii) actuated wheels to perform purely wheeled locomotion with differential steering; (iii) capability of operating after an overturn, due to the main the main body body symmetry. symmetry. Regarding the Regarding thewheeled wheeledlocomotion, locomotion,there there are some are some similarities with similarities the the with Kylin robot Kylin [28]: robot Two actuated wheels are placed at the end of rotating arms, and passive [28]: Two actuated wheels are placed at the end of rotating arms, and passive omni wheels omni wheels are adopted. The advantages of the WheTLHLoc with respect to are adopted. The advantages of the WheTLHLoc with respect to the Kylin are: (i) The the Kylin are: (i) The possi- bility of using possibility the longer of using legs, also the longer legs,inalso combination with the in combination with tracks, to overcome the tracks, obstacles; to overcome ob- (ii) the capability of operating after an overturn due to the symmetry; stacles; (ii) the capability of operating after an overturn due to the symmetry; (iii) the (iii) the most compact most and robust compact and design. robust design. The capability of The capability of operating operatingafterafterand andoverturn overturnisisinincommon commonwith with the thehybrid hybrid WT WTrobot ro- discussed in [29], but in comparison to this one, the following benefits bot discussed in [29], but in comparison to this one, the following benefits of WheTLHLoc of WheTLHLoc can be outlined: can (i) The be outlined: capability (i) The of using capability legs,legs, of using wheels and and wheels tracks in combination tracks in combination to overcome to over- obstacles and climb stairs (Figure 2c); (ii) the polygon of the contact come obstacles and climb stairs (Figure 2c); (ii) the polygon of the contact points points for the wheeled for the locomotion is much larger, allowing higher stability during turns, acceleration and breaking, wheeled locomotion is much larger, allowing higher stability during turns, acceleration and consequently higher motion performance. and breaking, and consequently higher motion performance. 3. Step and Stair Climbing and Descent 3. Step and Stair Climbing and Descent As already stated, the overall size of the robot is quite compact, since the considered As already payload stated, the is represented by overall cameras size andof environmental the robot is quite compact, sensors for since the considered surveillance and in- payload is represented by cameras and environmental sensors for surveillance spection. Moreover, a small robot can travel in narrow spaces that cannot be explored and in- by spection. humans and Moreover, is muchamore smallportable. robot canOntravel in narrow the other hand, spaces that cannot a drawback of smallberobots explored by is their capability of facing stairs. The climbing of steps and stairs can be performed according to the sequence rep- resented in Figure 3 (frames a to m). In case of single steps, the sequence a–m is per- formed once, while it is repeated n times in case of n steps.
sary to switch direction. Once the central zone of the tracks is in contact with the step edge (frame d), the legs rotate backwards until they touch the ground behind the robot (frame e) and completely lift the robot over the step (frames f to h); in this phase, both tracks and wheels rotate, pushing the robot forward. Once the robot center of mass is sufficiently beyond the step Machines 2021, 9, 10 5 of 11 edge, the robot moves forward on the tracks while the legs rotate forward (frame i–m) to face the next step, if present, or to complete the task. Figure 3. Stair-climbing sequence: the Figure 3. Stair-climbing legs lift sequence: theup theliftrobot legs body up the (a–c), robot bodythen rotate (a–c), then backward to complete rotate backward to the step complete climbing (d–h); then thethe stepgoes robot climbing (d–h); forward then the by tracks robot (i–l) goesthe to face forward by (m,n). next step tracks (i–l) to face the next step (m–n). After approaching the step, the rotation of the legs lifts up the robot body (frames a–c); The successful note let us that the execution robot, of this whichisisnot sequence not fully symmetric assured in any dynamic with and respect to the yz plane, friction must condition, since theface the step friction with and of tracks the rear side, wheels otherwise with in the the ground is following necessary phase (frame d) it would to complete the maneuvernot be possible in the to use the traction phase corresponding to the of the tracks frames d to g;due to the the therefore, presence motionofstrat- the omni wheels. Therefore, egy must be planned andif verified the robotbyapproaches the stairor analytical approach moving forward, numerically. a 180◦ 4, In Section pivoting the must be performed by differential steering before multibody simulation of the stair climbing maneuver is discussed.climbing the first step, imposing an equal speed to the two tracks with opposite directions. This maneuver The stair descent can be executed inverting the stair climbing sequence, as shown in can be used whenever it is necessary to switch direction. Figure 4. In the descent, the successful execution of the maneuver is less critical, since it is Once the the not necessary to overcome central zoneforce gravity of the bytracks meansisofinthe contact with traction ofthe step and wheels edgetracks. (frame d), the legs rotate backwards until they touch the ground behind the robot (frame e) and completely lift the robot over the step (frames f to h); in this phase, both tracks and wheels rotate, pushing the robot forward. Once the robot center of mass is sufficiently beyond the step edge, the robot moves forward on the tracks while the legs rotate forward (frame i–m) to face the next step, if present, or to complete the task. The successful execution of this sequence is not assured in any dynamic and friction condition, since the friction of tracks and wheels with the ground is necessary to complete the maneuver in the phase corresponding to the frames d to g; therefore, the motion strategy must be planned and verified by analytical approach or numerically. In Section 4, the multibody simulation of the stair climbing maneuver is discussed. The stair descent can be executed inverting the stair climbing sequence, as shown in Figure 4. In the descent, the successful execution of the maneuver is less critical, since it is not necessary to overcome the gravity force by means of the traction of wheels and tracks.
hines 2021, 9, x FOR PEER REVIEW 6 of 11 es 2021, 9, xMachines FOR PEER2021, 9, 10 REVIEW 6 of 11 6 of 11 Figure 4. Stair-descent sequence: the robot approaches the descent by tracks (a–b) and uses the legs to come Figure 4. Stair-descent downthe sequence: gradually without shocks robot approaches until by the descent thetracks tracks(a,b) touch theuses and nextthe step (c–f); thendown the gradually Figure legs 4. Stair-descent rotate backwardssequence: the robot (g) to complete theapproaches step descentthe(h–i). descent by tracks (a–b)legs andtouses comethe without shocks legsuntil the tracks to come downtouch the next gradually step (c–f); without shocksthen thethe until legstracks rotatetouch backwards (g)step the next to complete thethe (c–f); then step descent (h,i). legs4.rotate backwards Multibody (g) to complete Simulation theClimbing of Stair step descent (h–i). 4. Multibody Simulation of Stair Climbing The feasibility Theoffeasibility 4. Multibody Simulation theofstair Stairclimbing ofClimbing maneuver the stair climbingdiscussed maneuverindiscussed Section 3 in has been checked Section 3 has been checked with a dynamic model with ofa dynamicimplemented in model implementedRecurdyn, a multibody in Recurdyn, software a multibody commonly software used commonly used The feasibility the stair climbing maneuver discussed in Section 3 has been checked to simulate to complex simulate mechanical complex systems mechanical involving systems a large number involving a of large components number ofand to components and with a dynamic model implemented in Recurdyn, a multibody software commonly used perform efficient contact analysis by means of penalty algorithms [36]. In particular, to perform efficient contact analysis by means of penalty algorithms [36]. In particular, these to simulate complex mechanical systems involving a large number of components and to features canthese be exploited features tocansimulate the complex be exploited to simulatedynamic behavior the complex of a tracked dynamic system, behavior of a tracked sys- perform efficient contact analysis by means of penalty algorithms [36]. In particular, these composed of a high tem, numberofofa bodies composed (sprockets, high number track (sprockets, of bodies links, and rollers) in contact track links, with in contact and rollers) features can be exploited to simulate the complex dynamic behavior of a tracked system, each other and withwith eachthe ground other and within configurations, the ground in which are highly configurations, variable which duringvariable are highly the during composed of a high number of bodies (sprockets, track links, and rollers) in contact with vehicle motion [37]. motion [37]. the vehicle each other and with the ground in configurations, which are highly variable during the Figures 5–7 Figures represent simulation 5–7 representresults related simulation to a stair results withto related 300 mm of a stair run300 with andmm 160 of run and vehicle motion [37]. mm of rise 160 (standard mm of risesize (standard for buildings). size forFigure 5 shows buildings). the 5angular Figure shows the position vs. position angular time vs. time Figures 5–7 represent simulation results related to a stair with 300 mm of run and 160 graph for wheels graph(red), tracks (red), for wheels (black)tracks and legs (green). (black) and These motion laws legs (green). Thesehave beenlaws motion de- have been mm of rise (standard size for buildings). Figure 5 shows the angular position vs. time signed to realize the climbing designed to realize sequence of Figure the climbing 3 and are sequence imposed of Figure to the 3 and areactuators imposedintothe the actuators graph for wheels (red), tracks (black) and legs (green). These motion laws have been de- simulation. in Thethelaws are characterized simulation. The laws by areconstant speedby characterized phases andspeed constant stop phases. phases Since and stop phases. signed to realize the climbing sequence of Figure 3 and are imposed to the actuators in the Since instantaneous instantaneous speed variations are speednotvariations are not feasible (they wouldfeasible (they require wouldacceleration), infinite require infinite accelera- simulation. The lawsconstant tion), are characterized speed by constant phases and stop speed phases phases have and linked been stop phases. by Since cubic splines to obtain constant speed phases and stop phases have been linked by cubic splines to obtain twice instantaneous twice speedcontinuously variations aredifferentiable not feasible (they motion would require laws without infinite acceleration), acceleration peaks. continuously differentiable motion laws without acceleration peaks. constant speed phases and stop phases have been linked by cubic splines to obtain twice continuously differentiable motion laws without acceleration peaks. Figure 5. Motion laws5.of Figure tracks,laws Motion wheels, and legs of tracks, (angular wheels, positions and legs vs.positions (angular time) imposed in imposed vs. time) the multibody in the simulation to realize the stairmultibody simulation climbing sequence of to realize Figure the stair 3; stop climbing phases sequence and constant of Figure speed phases3;are stop phases linked byand cubicconstant splines to avoid Figure speed5. phases Motionare laws of tracks, linked wheels, by cubic andtolegs splines (angular avoid positions acceleration vs. time) imposed in the peaks. acceleration peaks. multibody simulation to realize the stair climbing sequence of Figure 3; stop phases and constant speed phases are linked by cubic splines to avoid acceleration peaks.
nes 2021, 9, x FOR PEER REVIEW 7 of 11 Machines 2021, 9, 10 7 of 11 s 2021, 9, x FOR PEER REVIEW 7 of 11 Figure 6. Multibody simulation results of the stair climbing sequence of Figure 3: Horizontal (x) and vertical (z) displacements of the reference frame origin of the robot main body (Figure 1) and pitch angle θ of the robot main body. Figure 6 represents the horizontal and vertical displacements of the robot reference Figure frame6. origin Figure 6. Multibody Multibody simulation simulation results (Figure results of the 1) with stairof the to stair climbing respect climbing sequence a fixed ofsequence reference Figure 3:ofHorizontal frame Figure located3: at Horizontal (x) and the (x)edge vertical lower (z) displace- and ments of the of vertical reference (z) frame the first displacements steporigin of the (Figure of the 7), robot reference mainwith together frame bodythe origin (Figure of robot1)pitch the and pitch robot angle main angle body (Figure 1) and θ. θ of the robot main body. pitch angle θ of the robot main body. Figure 6 represents the horizontal and vertical displacements of the robot reference frame origin (Figure 1) with respect to a fixed reference frame located at the lower edge of the first step (Figure 7), together with the robot pitch angle θ. Figure 7. Multibody Figure 7.simulation Multibody results of the results simulation stair climbing sequence of the stair of Figure climbing 3: Trajectory sequence of Figureof3:reference frame Trajectory of origin of the reference robot main body frame (Figure 1). origin of the robot main body (Figure 1). Figure 7 shows Figure 6 represents the trajectory of thetherobot horizontal body frameand vertical displacements origin with respect to theof the robot reference fixed reference frame frame andorigin (Figure 1)positions six significant with respect of theto climbing a fixed reference process:frame c, e, h,located at the m, n, e’. Thelower edge of Figure 7. Multibody symbols the simulation of such first six positionsresults step (Figure of 7),the correspond stairtoclimbing together with the sequence the frame robot of Figure pitch symbols the3: stair ofangle Trajectory θ. climbingof se- reference quence frame origin of Figure of theposition 3.Figure The robot main 7 shows e’thebody is (Figure trajectory similar to e1). ofafter the climbing robot body frame one step,origin so thewith cyclerespect e–h– to the fixed m–n–e’ can be repeated to climb all the stair. The time values corresponding to these c, reference frame and six significant positions of the climbing process: sixe, h, m, n, e’. Figure 7 shows the trajectory of six thepositions robot body frame origin with respect to the of fixed positions are marked in the graphs of Figures 5 and 6. The simulations have validated thestair climbing The symbols of such correspond to the frame symbols the reference framesequence and six significant of Figure positions 3. The of thee’climbing position process: c, e,climbing h, m, n, e’. oneThe feasibility of the maneuver and can be used to tune is thesimilar to estrategy climbing after as a function step, of so the cycle symbols of such six positions e–h–m–n–e’ can correspond be repeated totothe frame climb all symbols the stair. of the The timestair climbing values se- corresponding to these the step geometry. quence of Figure 3. The position e’ is similar to e after climbing one step, so the six positions are marked in the graphs of Figures 5 and 6. The simulations have validated cycle e–h– m–n–e’ can be 5. Internal repeatedand to the feasibility Layout climb of the Embodimentallmaneuver theDesign stair. and The cantimebevalues corresponding used to tune the climbing to these six as a function strategy positions are marked in the graphs of the step geometry. of Figures 5 and 6. The simulations have validated the In this section, the embodiment design of the robot is discussed. Figure 8 shows the feasibility of the maneuver and can be used to tune the climbing strategy as a function of internal layout 5. of the robot Internal body;and Layout theEmbodiment labels of the main Design components are listed in Table 1. the step geometry. In this section, the embodiment design of the robot is discussed. Figure 8 shows the 5. Internal Layout and Embodiment internal layout of the robotDesign body; the labels of the main components are listed in Table 1. In this section, the embodiment design of the robot is discussed. Figure 8 shows the internal layout of the robot body; the labels of the main components are listed in Table 1.
nes 2021, 9, Machines 9, 10 2021,REVIEW x FOR PEER 8 of 11 8 of 11 Figurelayout Figure 8. Internal 8. Internal of thelayout robot:of(a) the robot: of Section (a)the Section mainofbody the main alongbody along the xz the(b) plane; xz upper plane; view (b) upper (see the list of the view (see the main components in Table 1). list of the main components in Table 1). Main robot Table 1. plates The upper and lower internal (UP components. and LP) constitute the structural frame of the robot, on which the supports of the transmission components (gearmotors, axes, bear- Component Symbol in Figure 8 Characteristics ings, and gears of tracks and wheels) are fixed. The plates support also the other devices Outer Case OC 3D printed kevlar-reinforced nylon hosted in the robot body: Controller, battery, PWM drives, and video cameras. Free space Upper structural plate UP 3D printed micro-carbon-fiber-filled nylon is available Lower structural plate in the front and rear LP sides to place additional components (e.g., environmental 3D printed micro-carbon-fiber-filled nylon sensors, Leg gearmotor extra batteries) based LM on the required robot task. The structural components Micromotors RH-158-2s-12V, 26 rpm (plates and supports) are 3DTM Track gearmotor printed in micro carbon fiber filled nylon, Micromotors while the33 RH-158-2s-12V, upper rpm Wheel gearmotor WM EMG 30, 12v, 170 and lower external cases are 3D printed in Kevlar-reinforced nylon for higher impact re- rpm Motor driver (PWM) sistance. PD L298N Dual H-Bridge motor controller Leg axis LA Stainless steel tube (∅10 mm, 1 mm wall) Track axis TA High carbon steel tube (C45, ∅6 mm) Table 1. Main robot components. Leg gear LG 20:14, M2 Track gear Component TG Symbol in Figure 8 18:12, Characteristics M1.5 Omni wheels group Outer Case OW OC 3D printed kevlar-reinforced mm Rotacaster 95A-50 nylon Controller C National Instruments MyRio 1900 Upper structural plate UP 3D printed micro-carbon-fiber-filled nylon Battery B HRB 50C-4S-RC-LiPo, 14.8V, 2200 mAh Lower structural plate LP 3D printed micro-carbon-fiber-filled nylon Front camera FC Yosoo OV7670-300KP-VGA Rear cameraLeg gearmotor RC LM Micromotors Yosoo RH-158-2s-12V, OV7670-300KP-VGA26 rpm Slip ring Track gearmotor SR TM Micromotors RH-158-2s-12V, 12 mm-12wires miniature 33slip rpmring Wheel gearmotor WM EMG 30, 12v, 170 rpm Motor driver (PWM) PD internal plates The upper and lower L298N(UPDualandH-Bridge motor controller LP) constitute the structural frame of Leg axis LA Stainless steel tube (∅10 the robot, on which the supports of the transmission components mm, 1 mm wall) (gearmotors, axes, bear- Track ings, axis and gears of tracks TA and wheels) High carbon steel tube (C45, ∅6 mm) are fixed. The plates support also the other devices Leg gear LG Controller, battery, PWM20:14, hosted in the robot body: M2and video cameras. Free space drives, Track gear TG 18:12, M1.5 is available in the front and rear sides to place additional components (e.g., environmen- Omni wheels group OW Rotacaster 95A-50 mm tal sensors, extra batteries) based on the required robot task. The structural components Controller C National Instruments MyRio 1900 (plates and supports) are 3D printed in micro carbon fiber filled nylon, while the upper and Battery B HRB 50C-4S-RC-LiPo, 14.8V, 2200 mAh lower external cases are 3D printed in Kevlar-reinforced nylon for higher impact resistance. Front camera FC Yosoo OV7670-300KP-VGA The wheel gearmotors (Figure 8, WM) are directly connected to the wheels, and the Rear camera RC Yosoo OV7670-300KP-VGA necessary power supply and encoder signals are transmitted by slip rings (Figure 8, SR), Slip ring SR 12 mm-12wires miniature slip ring allowing continuous rotation, if necessary. The MyRio-1900 board by National Instruments (USA) has been selected as the control system, since it can properly manage the motion The wheel gearmotors (Figure 8, WM) are directly connected to the wheels, and the control of the six axes with encoders (two legs, two tracks, and two wheels), the video necessary power supply and encoder signals are transmitted by slip rings (Figure 8, SR),
Machines 2021, 9, 10 9 of 11 camera signals, and the WiFi connection with a host computer; moreover, it is equipped with a three-axis accelerometer to acquire data regarding the robot motion, and can manage an Inertial Measurement Unit. 6. Conclusions and Future Work In this paper, the conceptual design of a novel small-scale leg-wheel-track hybrid locomotion robot, named WheTLHLoc, is presented. A comparison with respect to other hybrid mobile robots is outlined in Section 2, highlighting similarities and differences. The mechanical architecture of WheTLHLoc has been designed to combine the advantages of tracked locomotion on soft and yielding terrains and the ones of wheeled locomotion on flat and compact grounds. The switch between these two modes is realized by means of rotating legs carrying driving wheels. The robot is capable of performing step and stair climbing and descent by moving in proper combination legs, wheels, and tracks. The effectiveness of the climbing maneuver is not obvious due to the limited dimension of the robot with respect to step height. Therefore, the feasibility has been verified by numerical simulation using the multibody software Recurdyn. Variations of this maneuver can be used to overcome obstacles with different shape, not squared, using the same principle of lifting the side of the robot close to the obstacle by the legs, then advance using the tracks, and finally completing the robot lifting by rotating backwards the legs. The detailed embodiment design has been completed, as discussed in Section 5. In the following of the research, the first prototype of the robot will be realized for an experimental validation of the proposed hybrid locomotion system and its test in real surveillance and inspection tasks. Author Contributions: L.B. conceived the locomotion system and the robot architecture; L.B., S.E.N., P.B., and M.B. developed the detailed embodiment design; P.F. supervised the scientific methodology for the functional design and the multibody simulations; M.B. performed the multibody simulations; all authors have prepared the manuscript. All authors have read and agreed to the published version of the manuscript. Funding: This research received no external funding. Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not applicable. Data Availability Statement: Data sharing not applicable. No new data were created or analyzed in this study. Data sharing is not applicable to this article. Conflicts of Interest: The authors declare no conflict of interest. References 1. IFR International Federation of Robotics, World Robotics 2019, Industrial Robots and Service Robots. Available online: https: //www.ifr.org (accessed on 1 September 2020). 2. Quaglia, G.; Visconte, C.; Scimmi, L.S.; Melchiorre, M.; Cavallone, P.; Pastorelli, S. Design of a UGV powered by solar energy for precision agriculture. Robotics 2020, 9, 13. [CrossRef] 3. Mateo Sanguino, T.J. 50 years of rovers for planetary exploration: A retrospective review for future directions. Robot. Auton. Syst. 2017, 94, 172–185. [CrossRef] 4. Murphy, R.R. Rescue robotics for homeland security. Commun. ACM 2004, 47, 66–68. [CrossRef] 5. Chun, W.H.; Papanikolopoulos, N. Robot Surveillance and Security. In Springer Handbook of Robotics; Siciliano, B., Khatib, O., Eds.; Springer: Berlin/Heidelberg, Germany, 2016; pp. 1605–1626. 6. Tadokoro, S. Rescue Robotics. DDT Project on Robots and Systems for Urban Search and Rescue; Springer: London, UK, 2009. 7. Dubey, S. Robot Locomotion—A Review. Int. J. Appl. Eng. Res. 2015, 10, 7357–7369. 8. Bruzzone, L.; Quaglia, G. Review article: Locomotion systems for ground mobile robots in unstructured environments. Mech. Sci. 2012, 3, 49–62. [CrossRef] 9. Kececi, E.F.; Ceccarelli, M. Mobile Robots for Dynamic Environments; ASME Press: New York, NY, USA, 2015; ISBN 978-0791860526. 10. Chung, W.; Iagnemma, K. Wheeled robots. In Springer Handbook of Robotics; Siciliano, B., Khatib, O., Eds.; Springer: Berlin/Heidelberg, Germany, 2016; pp. 575–594.
Machines 2021, 9, 10 10 of 11 11. Kim, J.; Kim, J.; Lee, D. Mobile robot with passively articulated driving tracks for high terrainability and maneuverability on unstructured rough terrain: Design, analysis, and performance evaluation. J. Mech. Sci. Technol. 2018, 32, 5389–5400. [CrossRef] 12. Ottonello, G.; Berselli, G.; Bruzzone, L.; Fanghella, P. Functional Design of Elloboat, a Tracked Vehicle for Launching and Beaching of Watercrafts and Small Boats. In Proceedings of the 14th IEEE/ASME International Conference on Mechatronic and Embedded Systems and Applications MESA 2018, Oulu, Finland, 2–4 July 2018. [CrossRef] 13. Bruzzone, L.; Berselli, G.; Bilancia, P.; Fanghella, P. Design Issues for Tracked Boat Transporter Vehicles. In Advances in Mechanism and Machine Science. IFToMM WC 2019. Mechanisms and Machine Science; Springer: Cham, Switzerland, 2019; Volume 73, pp. 3671–3679. [CrossRef] 14. Vukobratovic, M.; Borovac, B. Zero-moment point-thirty five years of its life. Int. J. Hum. Robot. 2004, 1, 157–173. [CrossRef] 15. Manchester, I.R.; Mettin, U.; Iida, F.; Tedrake, R. Stable dynamic walking over uneven terrain. Int. J. Robot. Res. 2011, 30, 265–279. [CrossRef] 16. Altendorfer, R.; Moore, N.; Komsuoglu, H.; Buehler, M.; Brown, H.B., Jr.; McMordie, D.; Saranli, U.; Full, R.; Koditschek, D.E. RHex: A biologically inspired hexapod runner. Auton. Robot. 2001, 11, 207–213. [CrossRef] 17. Bruzzone, L.; Fanghella, P. Functional Redesign of Mantis 2.0, a Hybrid Leg-Wheel Robot for Surveillance and Inspection. J. Intell. Robot. Syst. Theory Appl. 2016, 81, 215–230. [CrossRef] 18. Quaglia, G.; Bruzzone, L.; Oderio, R.; Razzoli, R.P. EPI.Q mobile robots family. In Proceedings of the ASME 2011 International Me- chanical Engineering Congress and Exposition, IMECE 2011, Denver, CO, USA, 11–17 November 2011; Volume 7, pp. 1165–1172. [CrossRef] 19. Quaglia, G.; Butera, L.G.; Chiapello, E.; Bruzzone, L. UGV epi.q-mod. In Advances on Theory and Practice of Robots and Manipulators; ROMANSY 2014. Mechanisms and Machine Science; Springer: Cham, Switzerland, 2014; Volume 22, pp. 331–339. [CrossRef] 20. Boston Dynamics, Handle. Available online: https://www.bostondynamics.com/handle (accessed on 1 September 2020). 21. Kashiri, N.; Baccelliere, L.; Muratore, L.; Laurenzi, A.; Ren, Z.; Hoffman, E.M.; Kamedula, M.; Rigano, G.F.; Malzahn, J.; Cordasco, S.; et al. CENTAURO: A hybrid locomotion and high power resilient manipulation platform. IEEE Robot. Autom. Lett. 2019, 4, 1595–1602. [CrossRef] 22. Hirose, S.; Shirasu, T.; Fukushima, E.F. Proposal for cooperative robot Gunryu composed of autonomous segments. Robot. Auton. Syst. 1996, 17, 107–118. [CrossRef] 23. Lee, G.; Kim, H.; Seo, K.; Kim, J.; Sitti, M.; Seo, T.W. Series of multilinked caterpillar track-type climbing robots. J. Field Robot. 2016, 33, 737–750. [CrossRef] 24. Fujita, T.; Sasaki, T. Development of hexapod tracked mobile robot and its hybrid locomotion with object-carrying. In Proceedings of the 5th IEEE International Symposium on Robotics and Intelligent Sensors, IRIS 2017, Ottawa, ON, Canada, 5–7 October 2017; pp. 69–73. [CrossRef] 25. Babu, N.; Sujatha, S.; Narayanan, S.; Balamurugan, V. Novel Hybrid Leg-Track Locomotion Robot and its Stability Analysis Using a Unified Methodology. Procedia Comput. Sci. 2018, 133, 486–493. [CrossRef] 26. Rea, P.; Ottaviano, E. Design and development of an inspection robotic system for indoor applications. Robot. Comput. Integr. Manuf. 2018, 49, 143–151. [CrossRef] 27. Hirose, S.; Fukuda, Y.; Yoneda, K.; Nagakubo, A.; Tsukagoshi, H.; Arikawa, K.; Endo, G.; Doi, T.; Hodoshima, R. Quadruped walking robots at Tokyo Institute of Technology. IEEE Robot. Autom. Mag. 2009, 16, 104–114. [CrossRef] 28. Han, C.; Xu, Y.; Xu, X.; Zeng, Z.; Lu, H.; Zhou, Z. Remote control and autonomous driving: The system-wide design of a wheel-track transformable robot—Kylin Blaster. In Proceedings of the 2018 Chinese Automation Congress (CAC), Xi’an, China, 23–25 November 2018; pp. 3446–3451. [CrossRef] 29. Ben-Tzvi, P.; Saab, W. A hybrid tracked-wheeled multi-directional mobile robot, ASME. J. Mech. Robot. 2019, 11, 041008. [CrossRef] 30. Kim, J.; Kim, Y.G.; Kwak, J.H.; Hong, D.H.; An, J. Wheel & track hybrid robot platform for optimal navigation in an urban environment. In Proceedings of the SICE Annual Conference, Taipei, Taiwan, 18–21 August 2010; pp. 881–884. 31. Michaud, F.; Letourneau, D.; Arsenault, M.; Bergeron, Y.; Cadrin, R.; Gagnon, F.; Legault, M.A.; Millette, M.; Pare, J.F.; Remblay, M.C.; et al. Multi-modal locomotion robotic platform using leg-track-wheel articulations. Auton. Robot. 2005, 18, 137– 156. [CrossRef] 32. Hirose, S.; Fukushima, E.; Damoto, R.; Nakamoto, H. Design of terrain adaptive versatile crawler vehicle HELIOS-VI. In Pro- ceedings of the IEEE International Conference on Intelligent Robots and Systems, Maui, HI, USA, 29 October–3 November 2001; pp. 1540–1545. [CrossRef] 33. Luo, Z.; Shang, J.; Wei, G.; Ren, L. A reconfigurable hybrid wheel-track mobile robot based on Watt II six-bar linkage. Mech. Mach. Theory 2018, 128, 16–32. [CrossRef] 34. Gao, X.; Cui, D.; Guo, W.; Mu, Y.; Li, B. Dynamics and stability analysis on stairs climbing of wheel-track mobile robot. Int. J. Adv. Robot. Syst. 2017, 14, 1–13. [CrossRef] 35. Zhou, F.; Xu, X.; Xu, H.; Zhang, X. A multimodal hybrid robot with transformable wheels. In Proceedings of the 2017 IEEE International Conference on Real-Time Computing and Robotics, RCAR 2017, Okinawa, Japan, 14–18 July 2017; pp. 139–144. [CrossRef]
Machines 2021, 9, 10 11 of 11 36. Bilancia, P.; Berselli, G.; Bruzzone, L.; Fanghella, P. A practical method for determining the pseudo-rigid-body parameters of spatial compliant mechanisms via CAE tools. Procedia Manuf. 2017, 11, 1709–1717. [CrossRef] 37. Mo, H.; Jianzhong, S.; Zirong, L.; Zhuo, W. Trench-crossing capability analysis of a reconfigurable tracked mobile robot. In Proceedings of the International Conference on Intelligent Robotics and Applications, Shanghai, China, 10–12 November 2010; pp. 509–518. [CrossRef]
You can also read