Institut Pascal UMR 6602 UBP/CNRS/IFMA
IFMA
UBP
French Institute for Advanced Mechanics
Blaise Pascal University Clermont-Ferrand II
Designing Agile Mobile Robots For Industrial Applications
[email protected] IFMA Campus de Clermont-Ferrand / Les Cézeaux, B.P. 265 63175 AUBIERE Cedex FRANCE
J.C. Fauroux, Institut Pascal Clermont-Ferrand, France MODTECH 2013, Sinaia, Romania
1
Agile Mobile Robots Agile mobile robots...
• •
Problem Problemsetting setting Agility Agility Existing Existing robots robots Synthesis Synthesis
PoBot PoBot OpenWHEEL OpenWHEEL i3R i3R MiniFAST MiniFAST Conclusion Conclusion
Terrestrial vehicles & robots
• • •
Wheel / Leg / Crawlers / Tracks Wheeled robots prevail (excellent energetic efficiency) Lack of agility / blocked on obstacles
Hybrid locomotion with additional mobilities
• •
Combining the advantages of wheel and leg Other solution: deformable frame
Our objective: Improving locomotion ● ● ● ●
J.C. Fauroux, Institut Pascal Clermont-Ferrand, France MODTECH 2013, Sinaia, Romania
...can evolve in more varied environments than classical mobile robots ...and have better performances for the task
Wheeled robots That climb obstacles With a stable behaviour In various 1D and 2D environments 2
Existing Agile Mobile Robots Categories of Hybrid robots
Problem Problemsetting setting Agility Agility Existing Existing robots robots Synthesis Synthesis
PoBot PoBot
• • •
Wheels on legs vs. deformable frame Active / Passive Difficulties: stiffness, power, control
WorkPartner (230 kg, 1.4m long, 7km/h) 4 wheels on actuated legs (3 DOF per leg) Steering via central joint Many locomotion modes automation.tkk.fi
RobuROC 6 (150 kg, 1.5m long) Active deformable frame 3 tiltable axles with passive warping Able to turn on itself, climb obstacles www.robosoft.fr Shrimp Passive deformable frame 6 wheels on 2 // bogies and 1 front linkage Excellent climbing abilities but requires 6 wheels www.asl.ethz.ch
OpenWHEEL OpenWHEEL i3R i3R
HPI-Racing Maverick Scout 4WD Rock Crawler Agile RC car with passive spatial deformable frame http://www.hpiracing.net.cn
MiniFAST MiniFAST Conclusion Conclusion
J.C. Fauroux, Institut Pascal Clermont-Ferrand, France MODTECH 2013, Sinaia, Romania
Equipment deployment method and apparatus Spittle et al., US 2003/0188416A1, 2003 Specific pole climbing robot with passive rollers on arms
3
Synthesis of Agile Mobile Robots A similar design approach based on structural synthesis
Problem Problemsetting setting Agility Agility Existing Existing robots robots Synthesis Synthesis
PoBot PoBot OpenWHEEL OpenWHEEL i3R i3R MiniFAST MiniFAST Conclusion Conclusion
J.C. Fauroux, Institut Pascal Clermont-Ferrand, France MODTECH 2013, Sinaia, Romania
Expressing the problem - Define the task - Define the required components - Elements C in contact with the environment (wheels) - Main chassis / frame F - Payload... - Define the relative motions between F and C - Translation / Rotation / Other - Fix direction (X/Y/Z) / Other - Degree of coupling between motions Solving the structural synthesis problem - Choose a generation strategy - At random - By substitution - Constraint based - By inference - Enumerative - Evolutionist ... Synthesis - Generate kinematics that can produce loop - exact motions... - … or approximate motions → evaluate if acceptable Towards dimensional synthesis...
Objectives of this work
•
• • • •
Formalize a design approach for agile mobile robots Illustrate it on three design projects : Pobot OpenWHEEL MiniFAST
4
PoBot: A Pole Climbing Robot Expressing the problem into constraints
Problem Problemsetting setting PoBot PoBot Constraints Constraints
• • • • • • •
C1. Design a robot that can climb along... C2. ...and rotate around poles (1D environment) C3. Tangential obstacles should be crossed C4. No energy to maintain the robot statical on the pole C5. Cylindrical and conical poles (Diam 100-300 mm) C6. Robot specifications: cube 500 mm, 10kg, climbing speed 50 mm/s C7. Payload dimensions: cube 100 mm, 1kg
Self-locking Self-locking Synthesis Synthesis
OpenWHEEL OpenWHEEL i3R i3R
•
MiniFAST MiniFAST
•
Conclusion Conclusion
J.C. Fauroux, Institut Pascal Clermont-Ferrand, France MODTECH 2013, Sinaia, Romania
Z
Self locking A solution for C4 → Self-locking Def: phenomenon where locking is obtained only by friction and whatever the intensity of external forces
Punctual contact P2
G
Punctual contact P1
Pole
Weight P O
5
PoBot: Rolling Self-Locking Alternative motion ● ●
Problem Problemsetting setting PoBot PoBot
Two self-locking frames connected by a contracting mechanism Complex + jerky motion
C1 → Continuous ascension motion ● ●
Locating the contact points directly on rollers Simpler + continuous motion
Z
Constraints Constraints Self-locking Self-locking Synthesis Synthesis
Roller R2
OpenWHEEL OpenWHEEL i3R i3R MiniFAST MiniFAST Conclusion Conclusion
G
Remarks ●
Actuators:
Roller R1
Weight P
- 2 rollers - 1 roller (the one closest to heavy parts) → R1 J.C. Fauroux, Institut Pascal Clermont-Ferrand, France MODTECH 2013, Sinaia, Romania
●
Pole
What is the condition for self-locking ?
O
6
Rolling Self-Locking Condition Static equilibrium ●
Problem Problemsetting setting PoBot PoBot Constraints Constraints
Momentum expressed in C1
Non-slipping condition ●
Expressed in C1 ,
●
Only roller R1 propels, R2 is free
●
With µ the friction coefficient
OpenWHEEL OpenWHEEL i3R i3R MiniFAST MiniFAST Conclusion Conclusion
N 1≥m g/
(1) + (3) →
amg N 1= b tan
(5) + (6) →
b tan a≥
with =arccos d / b ● J.C. Fauroux, Institut Pascal Clermont-Ferrand, France MODTECH 2013, Sinaia, Romania
●
T 1≤ N 1
(4)
(2) (3)
b
a
Self-locking condition (2) + (4) →
(1)
d
Self-locking Self-locking Synthesis Synthesis
N 1 =N 2 T 1=mg m g a cos =b sin N 2
N2
T1
C2
(5) N1
(6) G
(7)
Weight mg
Roller R1
Roller R2
C1
Pole z O
(7) does not depend on mass, only on geometry and friction If θ → 0 then N1 → ∞ (but stiffness is not infinite)
y
7
Pobot: Structural Synthesis C2 → additional mobilities
C5 → adjustable arms ●
Problem Problemsetting setting
C1
G
PoBot PoBot
●
C2
Roller R
Spherical joint S
Pole
●
Turret T
Constraints Constraints
Weight P
C3 → symmetrical splitting
MiniFAST MiniFAST
Turret T
1
W
S
Fs
Robot G
Minimum diameter
C1
Spring 2ns C1
Structural synthesis
Intermediate diameter
C21
C22
Pole C22
J.C. Fauroux, Institut Pascal Clermont-Ferrand, France MODTECH 2013, Sinaia, Romania
C21
Spring 1 joint S Spherical
Roller R
Conclusion Conclusion
Maximum diameter
Structural synthesis
U
OpenWHEEL OpenWHEEL i3R i3R
Pressure angle δ should be 90° Translation of contact points approx. by revolute joints M-N Dim. synthesis for M-N location
Structural synthesis
Self-locking Self-locking Synthesis Synthesis
●
Support triangle C1 C21 C22
Spherical joint S2
Dimensional synthesis
8
PoBot: Force Regulation Linkage Sub-constraints for the force regulation linkage
Problem Problemsetting setting PoBot PoBot Constraints Constraints Self-locking Self-locking
• • •
MiniFAST MiniFAST
•
•
Conclusion Conclusion
J.C. Fauroux, Institut Pascal Clermont-Ferrand, France MODTECH 2013, Sinaia, Romania
C5.2 Better without actuator C5.3 Constant pushing forces FC and F'C for all diameters
Linkage synthesis
Synthesis Synthesis
OpenWHEEL OpenWHEEL i3R i3R
C5.1 Coupled symmetrical motions of C21 and C22 with respect to ∆
• •
C5.1 → symmetrical folded levers UMC21 & VNC22 C5.2 → springs → traction force FS linearly depends on distance WS → contradicts C5.3
Maximum diameter
U
Structural synthesis C21
Spring 1 W
S
Spring 2ns
∆
Minimum diameter
C1
Fs
Intermediate diameter
C22
Dimensional synthesis
C5.3 → Additional levers US VS → Non linear relation F C= f(WS) Singularity: UVS aligned → stiffness F C / WS = structural stiffness
9
PoBot: Experimental validation CAD model and prototype
•
Problem Problemsetting setting
•
PoBot PoBot
Tested on a conical wood pole (low friction, µ = 0.47), height 8m, diam. 210 / 140 mm Cannot climb without the force regulation linkage
V
Constraints Constraints
eo d i
Self-locking Self-locking Synthesis Synthesis
OpenWHEEL OpenWHEEL i3R i3R MiniFAST MiniFAST Conclusion Conclusion
J.C. Fauroux, Institut Pascal Clermont-Ferrand, France MODTECH 2013, Sinaia, Romania
Conclusion on the agile pole climbing robot Pobot
• • • •
New concept of rolling self-locking thanks to additional mobilities An agile climbing robot that adapts to many conical poles One solution was synthesized based on design constraints Joint PCT patent IFMA / Thales FR2929228 WO2009118409 10
OpenWHEEL: Crossing obstacles with four wheels only Expressing the problem (constraints + min function) Problem Problemsetting setting PoBot PoBot OpenWHEEL OpenWHEEL i3R i3R Constraints Constraints Synthesis Synthesis Locomotion Locomotion
MiniFAST MiniFAST Conclusion Conclusion
• • • • • •
C1. Using only 4 wheels C2. No overconstraint on irregular 2D grounds C3. Changing direction with steering C4. Can cross obstacles C5. Stability during obstacle crossing F1. Minimize the number of joints / actuators
The OpenWHEEL family of agile mobile robots
• • •
C4 → The OpenWHEEL family of robots with additional mobilities Sas suspension mechanisms / Ia inter-axle mechanisms F1 → factorizing mobilities within Ia avoids to include them into each Sas Wireless connection A3 Rear
J.C. Fauroux, Institut Pascal Clermont-Ferrand, France MODTECH 2013, Sinaia, Romania
Wheel W31
S 32
Control
S 31
CAN Bus
W 32
A2
W22 S 22
Control
I2 W21
S 21
W11
S 12
A1 Control
I1
W12
Camera
S 11
nt Fro Z X Y
11
OpenWHEEL: Synthesis of the central mechanism Synthesis of the basic mobile platform Problem Problemsetting setting
• •
C1 → 4 wheels like most vehicles C2 → OpenWHEEL i3R → one central passive warping revolute joint R0 around x + 2R steering passive joints → OpenWHEEL 4sRR → variant with 4R joints
PoBot PoBot OpenWHEEL OpenWHEEL i3R i3R Constraints Constraints Synthesis Synthesis
• •
F1 → R0 replaces four suspensions mechanisms → i3R better than 4sRR C3 → OpenWHEEL i3R → double Ackermann around R1 R2 (4 modes of steering) and stability unchanged → OpenWHEEL 4sRR → decrease in stability during steering
Locomotion Locomotion
R0
MiniFAST MiniFAST Conclusion Conclusion
R22
R1
R2
R0
R12
R21 R11
W22
W22
W12
Z
Y
W21 J.C. Fauroux, Institut Pascal Clermont-Ferrand, France MODTECH 2013, Sinaia, Romania
W11
OpenWHEEL i3R [IROS 06]
W21
W11
OpenWHEEL 4sRR [CMSM 07]
W12
X
12
OpenWHEEL: Concept of Exploring Wheel Re-using the mobilities on flat ground to locate an “Exploring Wheel” over the obstacle
Exploring wheel (W12)
Wheel (W12) joint R12 Problem Problemsetting setting PoBot PoBot OpenWHEEL OpenWHEEL i3R i3R
Rear frame (F2) Rear steering joint R2
Constraints Constraints
A22 O22
Synthesis Synthesis W22
Locomotion Locomotion
P22 Rear axle steering angle 2
J.C. Fauroux, Institut Pascal Clermont-Ferrand, France MODTECH 2013, Sinaia, Romania
x12 A12
O12 zF2
B12 zF1
O
B22
P11
A21
G'
W21
P21 rW O0
x0
Step Obstacle
Advantages
O21
y0
hS
Lifting polygon Stability on three wheels
B21
z0
O11 W11
xA2 H
Reconfigured rear axle (A2)
A11
G1 B11
G
G2
OA1
2
Front steering joint R1 Exploring front axle (A1)
2 =O F
F1
T2 yA2
x
= xF
F1
Front axle steering angle 1
xA1
yA1
yF2
zA2
zA1 T1
W12
OA2
MiniFAST MiniFAST Conclusion Conclusion
y12
Warping angle 0 Central warping joint R0
Front frame (F1)
z12
C4 → R0 lifts the exploring wheel C4 → R1 brings the exploring wheel forward C4 → Front-Rear + Left/Right 13 symmetries → 4 exploring wheels
OpenWHEEL: Obstacle Crossing and Stability Checking stability on three wheels
PoBot PoBot OpenWHEEL OpenWHEEL i3R i3R
●
Constraints Constraints Synthesis Synthesis Locomotion Locomotion
MiniFAST MiniFAST Conclusion Conclusion
●
Wheel W11 (front-right)
Wheel W12 (front-left)
Wheel W21 (rear-right)
1)
2)
3)
Front axle steering
Problem Problemsetting setting
2D approximate model Stable if the lifted wheel is inside the turn Climbing process in 19 stages and 6 phases
G
4)
G
G
W22
G
W21 W11
G
6)
W12
7)
G
8)
G
G W21
Stable J.C. Fauroux, Institut Pascal Clermont-Ferrand, France MODTECH 2013, Sinaia, Romania
W12
Wheel W22 (rear-left)
W11
5)
Rear axle steering
●
Unstable
Stable
W22
Unstable
14
OpenWHEEL i3R locomotion mode A - Prepairing
1
W22
W12
2
3
4
5
7
8
9
12
13
14
17
18
19
W11
W21
Problem Problemsetting setting Wheel center motion
PoBot PoBot
B - W11 climbing
6
C - W12 climbing
Wheel lifting Wheel landing
OpenWHEEL OpenWHEEL i3R i3R
Support polygon For a very stable configuration (Four contact points)
For a stable configuration (Three contact points)
Constraints Constraints Synthesis Synthesis
10
D – Going forward
11
E - W21 climbing
Locomotion Locomotion
MiniFAST MiniFAST Conclusion Conclusion
J.C. Fauroux, Institut Pascal Clermont-Ferrand, France MODTECH 2013, Sinaia, Romania
15
F - W22 climbing
16
G - Conclusion
15
OpenWHEEL: Experimental validation Modeling and testing ● ●
Problem Problemsetting setting
●
PoBot PoBot
2D simplified model in top view 3D multi-body model (Adams) Demonstrators at several scales
V
eo d i
24V actuator (330W)
OpenWHEEL OpenWHEEL i3R i3R Constraints Constraints
Dual-stage 10.9x chain transmission
Synthesis Synthesis ATV tire
Locomotion Locomotion
MiniFAST MiniFAST Conclusion Conclusion
Conclusion on OpenWHEEL ●
●
● J.C. Fauroux, Institut Pascal Clermont-Ferrand, France MODTECH 2013, Sinaia, Romania
Synthesis of OpenWHEEL i3R: an agile mobile robot with deformable frame Can climb 2/3 of centre of mass altitude with only 4 wheels Minimizing actuator number: only 1 central actuator for moving 4 exploring wheels
16
MiniFAST: A new suspension for fast obstacle crossing Expressing the problem Problem Problemsetting setting
• • •
Dynamic obstacle crossing with 4-wheeled vehicles High speed or/and High obstacle → Pitch over How to improve the crossable height or/and speed without pitch over ?
PoBot PoBot OpenWHEEL OpenWHEEL i3R i3R MiniFAST MiniFAST Problem Problem Constraints Constraints Synthesis Synthesis
Conclusion Conclusion
J.C. Fauroux, Institut Pascal Clermont-Ferrand, France MODTECH 2013, Sinaia, Romania
17
MiniFAST: Requirements Expressing the problem with constraints Problem Problemsetting setting PoBot PoBot OpenWHEEL OpenWHEEL i3R i3R MiniFAST MiniFAST Problem Problem Constraints Constraints
• • • • • • •
C1. The wheel must have a damping vertical translation TZ C2. The wheel must have a damping horizontal translation TX C3. The wheel has a steering rotation RZ around the vertical axis Z C4. The wheel has a driving rotation RY applied by an engine C5. All the motions are decoupled C6. Good stiffness KX C7. Good stiffness KY
RZ steering
Z
G
Synthesis Synthesis
Conclusion Conclusion
RY driving X
J.C. Fauroux, Institut Pascal Clermont-Ferrand, France MODTECH 2013, Sinaia, Romania
TZ damping
TX damping
N 18
MiniFAST: Linkage synthesis V1
- Decreases the nonsuspended mass with respect to V2
120
101
130
150 100
V2'
255 252
110
PoBot PoBot
Synthesis Synthesis
Z X
Conclusion Conclusion
V1 V2 V2' V3
V2 254 241 243 242
200
Z X
245 255
Cylinder 2
J.C. Fauroux, Institut Pascal Clermont-Ferrand, France MODTECH 2013, Sinaia, Romania
241
TZ TX RZ RY Decoup. KX KY
Y
Cylinder 1
Constraints Constraints
243 244
MiniFAST MiniFAST Problem Problem
254
X
244 251 253 252 220
210
C1 C2 C3 C4 C5 C6 C7
OpenWHEEL OpenWHEEL i3R i3R
245
200
Z
Cylinder 1
- Risk of bending at longitudinal 111 shock
251
242
140
Problem Problemsetting setting
Cylinder 2
244
253
220
210
V3 300
- Too many prismatic joints (expensive, butting) - Low horizontal limb - Bad lateral guidance (KY)
354
355 345 351
- Best KX stiffness because works in pure compression (no bending) - But coupling appears
341 353
343 352
342 320
Z X
310
19
MiniFAST: Linkage synthesis V4 400
Problem Problemsetting setting
440 Z2
450
410
PoBot PoBot OpenWHEEL OpenWHEEL i3R i3R
O X
MiniFAST MiniFAST
Z Y
V5 500
Synthesis Synthesis
Conclusion Conclusion
X
540
640 Z1 Z2
Z 633 631
X
Y
510
600 632 610 650
V4 V5 V6 V8
V7 X
Z 740
Y
Z
Z2
Z1
Y
620
661
C1 C2 C3 C4 C5 C6 C7
433 434 420 421 431 432 470
Problem Problem Constraints Constraints
V6
TZ TX RZ RY Decoup. KX KY
- Local decoupling - Collision during steering - Wheel attachment problem
- Remaining coupling Tx-Rz - Wheel attachment problem
710 720 733
700
- Easier wheel attachment - No steering
731
E E731 733
E770 550 J.C. Fauroux, Institut Pascal Clermont-Ferrand, France MODTECH 2013, Sinaia, Romania
531 532 533 534 570
- Remaining coupling TxRz
E732
761
770
732 750
20
840
800
C1 C2 C3 C4 C5 C6 C7
TZ TX RZ RY Decoup. KX KY
MiniFAST: Final solution
V9 Z1
Problem Problemsetting setting
V9
854 853 852 852' 850
PoBot PoBot OpenWHEEL OpenWHEEL i3R i3R MiniFAST MiniFAST Problem Problem
Conclusion on MiniFAST ●
Constraints Constraints Synthesis Synthesis
● ●
Conclusion Conclusion ● ●
J.C. Fauroux, Institut Pascal Clermont-Ferrand, France MODTECH 2013, Sinaia, Romania
●
Synthesis of a suspension for agile mobile robots and dynamic obstacle crossing 4 mobilities, including TX damping Spherical translations approximate linear translations De-coupling is achieved in the ref. position Structural synthesis based on constraint analysis : 9 solutions, 3 2D, 6 3D, 8 parallel and 6 spatial kinematics PCT patent IFMA FR2980398 and WOxxxx
Z2
861 833 832 870 831 Z Y 810 820
V
X
eo d i
21
Conclusion / Future work Main results ●
●
3 innovative agile mobile robots were synthesized for pole climbing and stable obstacle crossing at low/high speed Synthesis method based on the expression of constraints on the relative mobilities between the wheels and the robot frame 3. MiniFAST agile mobile robot for obstacle crossing at high speed
Problem Problemsetting setting PoBot PoBot OpenWHEEL OpenWHEEL i3R i3R MiniFAST MiniFAST
1. PoBot agile pole climbing robot
Conclusion Conclusion
2. OpenWHEEL agile mobile robot for obstacle crossing at low speed
Future work ● J.C. Fauroux, Institut Pascal Clermont-Ferrand, France MODTECH 2013, Sinaia, Romania
●
Systematic exploration of design space Kinematic optimization
Acknowledgement ● ● ●
French National Research Agency (ANR) FEDER Europe in Auvergne All the involved IFMA students
22