Crawling, walking or rolling for obstacle-crossing - Jean-Christophe

Hybrid locomotion combines wheels and active legs or frame. Crawling ... OpenWHEEL i3R: a hybrid mobile robot with deformable frame ... phases A-G. Phases ...
2MB taille 2 téléchargements 295 vues
58 Crawling, walking or rolling for obstacle-crossing ? Bio-inspiration for the OpenWHEEL i3R agile mobile robot [email protected]

[email protected]

Clermont University – French Institute for Advanced Mechanics (IFMA) LaMI – B.P. 10448 – 63000 CLERMONT-FERRAND, France

Clermont University – Blaise Pascal University (UBP) LIMOS – B.P. 10448 – 63000 CLERMONT-FERRAND, France

Paradigm for stable climbing

OpenWHEEL i3R OpenWHEEL

- Crawling / Walking is used by most terrestrial animals : efficient in natural environment with irregular and unstructured ground but slow, difficult to control, and requires a lot of energy. - Wheels are used on most human vehicles : energy-efficient and fast but only on smooth and hard terrain. - Hybrid locomotion combines wheels and active legs or frame

Parallel Mechanism

Innovative Mechanism

OpenWHEEL i3R is an agile mobile robot with low actuation (only one central active joint, passive steering induced by wheel actuation) and multiple modes of locomotion for rolling on smooth surfaces or crossing obstacles

Inter-Axle Mechanism

WorkPartner (230 kg, 1.4m long, 7km/h) 4 wheels on legs (3 DOF per leg) automation.tkk.fi [Halme 2003]

I

Wireless connection

Data Bus

A1 ar Re Control Nomad Dual Ackermann steering strategy www.frc.ri.cmu.edu/projects/lorax [Rollins 1998, Apostopoulos 2001]

Roller-Walker (24kg, 0.5m long) Convertible wheels / actuators only on legs Dual locomotion mode: walking / roller-skating www-robot.mes.titech.ac.jp [Hirose 1996]

Serial Mechanism

S32

S22

A2 Control

I1

A3 Control

I2

S21

S31

Fro

nt

S11

Front frame (F1) Exploring wheel (W12) yW12

S11

Actuated Wheel Lama Peristaltic crossing of sandy areas www.laas.fr [Lacroix 2002]

- The present work aims at keeping the interesting obstacle crossing capacities of hybrid locomotion but replaces the multiple legs by a single deformable frame, that allows to decrease the number of actuators

Suspension Mechanism

Double Wishbone

1)

2)

3)

G

G

W 11

12V 48Ah traction battery

W22

zF2

yF2

24V actuator (330W)

8)

G

G W 21

Unstable

W 22

Stable

Unstable

T1 zA1 xA1

Front steering joint R1 Exploring front axle (A1)

OA1

0 zF1 xF1= xF2

G1

A11

B22

T2

G

W11

zA2 yA2

O11 hS

xA2 P11

H G'

A21

G2

Step Obstacle B21 O21

Reconfigured rear axle (A2)

Electromagnetic clutch

Climbing process

W21 P21 z0 y0

Curtis programmable DC controller

Dual-stage 10.9x chain transmission

rW

x0 O0

- Stability is evaluated by a simplified 2D model in top view (figure below, left column).

Wheel center motion

Wheel landing

- A 3D multibody model (Adams) confirms feasibility (figure below, rightcolumn) [Fauroux 2006]

Support polygon For a very stable configuration

ATV tire

(Four contact points)

For a stable configuration

(Three contact points)

W22

Phase A Preparing

W12

Vi

- 1.85m long, 1.38m wide, 0.98m high, 200kg - 24V DC actuators 330W with peak torque of 100Nm + external encoder + Two or four 12V 48Ah traction batteries - Central clutch → no overconstraint when rolling → no suspension required - Fast 45° central warping : only one second (15x faster than V1) - The same high level NXT controller as V2 generates 8kHz PWM Pulse Width Modulated motor signals - Used as an average tension to control low level Curtis DC controllers (24V, 70A)

Phase B W 11 climbing

2

OpenWHEEL i3R V2

3

Rotation sensor

9V 8 Ah battery

4

Motor multiplexer

5

Phase C W 12 climbing

Wifi remote

6

Control unit ARM 48 MHz 256 kB RAM

9V actuator (5W)

CAD model

Overconstrained zero clearance twin transmission Contact sensor

- Small scale (50cm) with closed loop control with access to PID parameters - Powerful 9V actuators (5W) and many sensors : contact, steering angle, distance (US), 3D accelerometer to measure tilting - When not horizontal, steering reconfiguration generates induced axle warping - Solution: unwarp the central joint / declutch the central joint

06

06

Without warping correction

With warping correction

7

- The whole robot V3 - Detail of the warping system : - reduction ratio 10.9 - max torque > 1000Nm - electric clutch

8

9

Steering-warping Coupling equation [Fauroux 2010]

sin 0 =−tantan 2 

Phase D Going forward

Front frame (F1) is supposed fixed to the ground and submitted to a roll angle α and a pitch angle β

10

Phase E W 21 climbing

11

OpenWHEEL i3R V1

Bio-inspiration from balancing of walking natural creatures

12

OpenWHEEL i3R V1, a reduced model built with the Mindstorms RCX robotics kit - 1.5kg autonomous robot - The first to validate the autonomous climbing process - Open loop (no coders in the motors) - 9V DC actuators (1.1W) - Two controllers synchronized by IR message exchange - Best obstacle height : 55 mm, 67% of the height of the centre of mass

- Longitudinal balancing to increase stability: walking uphill induces for bipeds an increasingly flexed posture of hip, knee and ankle at initial foot contact as well as a progressive forward tilt of pelvis and trunk [Leroux 2002] - Solution : mounting batteries or payload on a longitudinal slider

[Muybridge 1883]

G1 170 mm

- OpenWHEEL i3R: a hybrid mobile robot with deformable frame - Can climb obstacles as high as 2/3 of the altitude of its centre of mass with only 4 wheels - Only 1 supplemental actuator → minimal actuation, good stiffness - 3 small and full scale implementations / 1 control architecture - Problems & solutions : front-rear non-symmetry, steering-warping coupling

13

G G2

W12 17 5  m

m

0  19

W22

m m

14

Front-rear non symmetry [Fauroux 2008]

Phase F W 22 climbing

- Axle A1 climbs easily whereas axle A2 has difficulties - Solved by adding a counter-weight CW of 150 g - Non-symmetry was not predicted because of 2D approximation - On flat ground, the 2D model is exact - With some pitch: - In 2D, stability margin P 2 G ' =b cos /2

- Lateral balancing during obstacle crossing: centre of mass of bipeds is moved laterally by combined rolling of hip and ankle [Hof 2007] - For OpenWHEEL, lateral balancing comes from steering reconfiguration (stages 3, 7, 12 and 17) - Complementary solution : lateral sliding of payload

15

P 2 G ' =b cos /2−hl hs /b

- In 3D

16

- Stability is favoured at stages 3 and 7 and penalized at stages 12 and 16 - Adding CW equilibrates the climbing capacities of front and rear axles

- The current obstacle-crossing process of OpenWHEEL i3R is very close quadruped locomotion (example of the cat) a)

2D model: On flat ground

b)

d)

2D model: On a step

3D model: stability margin nullifies

b G hl

- Improving volume for payload → Replacing the inter-axle mechanism by an exoskeleton, such as for arthropods

O0

rW

G' x0

P2

P1

 P1

z0 x0 P2

O0

- Towards new robots & vehicles for agile all-terrain mobility & clearance performance

c)

3D model with counterweight CW: stability is back

hs

G'

18

3D model: On a step

e)

f)

G'

Phase G Conclusion

G'

 G 

CW

hl z0

The authors wish to thank the following students from IFMA and UBP for their contribution to this work : Guillaume MALVAL, Emilie PORTALES, Morgann FORLOROU, Sylvain MARCO, Yohann KOBERLE, Damien ROUX, Laurent GENEVAY, Michel TOU, Fabien BIANCHINI, Romain CARTAILLER, Farhat HZAG, Sylvain METAIS, Christophe NOELLAT.

O0

 P1

rW x0

17

G'

G

z0

Example: Stage 12

2D model: supposed stable

P2

G'

19

W21

W11

High

Low

1

Sensor multiplexer

- The obstacle-crossing mode was created by interpolating between several stable configurations where the robot lays either on three or four contact points. - The complete process has 19 stages divided into 7 phases A-G. Phases B-C and E-F concern the front and rear wheels respectively and have a similar structure: - Steering the other axle for stability - Lifting the exploring wheel above the obstacle - Bringing the exploring wheel forward - Landing the exploring on the obstacle

Wheel lifting

05

G

B11

OA2

P22

G

OF1=OF2

A22 O22

W 22

7)

W 12

Rear steering joint R2

Battery and payload front cradle

G

6)

G

yA1

Rear frame (F2)

OpenWHEEL i3R V3

4)

W 21

xW12 A 12

O12

Central warping joint R0

Wheel W 22 (rear-left)

W 11

B12

Innovative Suspension

W 12

Stable

W12

Swing arm

Wheel W 21 (rear-right)

zW12

Wheel (W12) joint R12

S

Wheel W 12 (front-left)

5)

The rover can climb on obstacles with only 4 wheels: 3 wheels for stable support of the robot (points P11, P21, P22) and 1 exploring wheel to crawl over the obstacle (wheel W12 on figure below)

Vision

Wheel W 11 (front-right)

Front axle steering

A family of rovers developed at LaMI / IFMA with articulated frame and / or innovative suspensions

Static stability on 3 wheels - Configurations 2) 4) 6) 8) are unstable - The stable configurations 1) 3) 5) 7) occur when the exploring wheel is located on the inside of the turn

Origin of the name i3R: - i for the inter-axle central mechanism - 3R for the number of revolute joints used in this mechanism - two passive steering rotations for front and rear axles - one central active warping joint

Rear axle steering

Wheeled / Legged / Hybrid Locomotion on unstructured terrain

o de

2

o de i V

1