Distributed Compliance Controllers for Legged- Robot with Geared Brushless DC Joints

Transcription

1 Distributed Compliance Controllers for Legged- Robot with Geared Brushless DC Joints Lan Yue Ji, Sebatian Bartsch, Frank Kirchner DFKI Bremen & Universität Bremen Robotics Innovations Center Director: Prof. Dr. Frank Kirchner

2 Overview General information of the SpaceClimber project Types of Mobility for Crater Exploration Technical Aspect of SpaceClimber Robot the Controller Design Actuator Estimation of the contact torque Control schematic Exemplar experimental result Conclusion and Outlook 2

3 Project SpaceClimber Goal: Development of a six-legged, biologically-inspired, energy efficient, and adaptable free climbing robot for extraterrestrial exploration The system has to be able to move freely and securely in crater environment cope with inclinations up to 80% navigate semi-autonomous carry a scientific payload Future space qualification has to be taken into account Funding: The project SpaceClimber is funded by the German Space Agency (DLR, Grant number: 50RA0705) and the European Space Agency ESA (Contract no.: 18116/04/NL/PA) 3

4 Mobility for crater exploration Multitude of concepts for robots that could provide mobility in crater environments (e.g. wheels, tracks, legged-wheels and robot teams) Walking systems offer the highest mobility Trajectories with partial ground contact (no bulldozing, negotiate obstacles) Variable selectable foothold position High traction in steep terrain High flexibility Various walking pattern and postures Omni directional walking Multi-functionality Scarab Rover (CMU) [1] Icebreaker (CMU) [2] TRESSA (JPL) [3] CESAR (Uni Bremen / DFKI) [4] 4

5 Evolution of DFKI Legged Robots 5

6 SpaceClimber Prototype Technical Data Size [LxWxH]: 820mm x 900mm x 220mm Weight: 23Kg Actuators (# = 25 x Body, 1 x Head) : Legs: 24x RoboDrive, Harmonic Drive 1:100 (6x Spring in lower legs) Body: 1 RoboDrive, Harmonic Drive 1:160 Head: 1x Dynamixel DX117 Sensors (# = 193) : Actuator: Position, speed, current, temperature, noise level Foot: 4x pressure, (4x DMS), 3 axes acceleration, temperature, piston immersion Leg mounting: 6 axes force-torque-sensor Body: IMU, overall power consumption Head: Laser range finder, CMOS camera Controller: Microblaze (51,96Mhz) on Spartan3A 1200 Communication: Wireless LAN (telemetry, commands), DECT (emergency switch) Power Supply : 4000mAh 6

7 Control Hierarchy of SpaceClimber 7

8 Control Frequencies 8

9 Actuator Design Selected components BLDC-motor: RoboDrive ILM 50x8 Gear: 100:1 HarmonicDrive Rotor position: Digital Hall sensors Absolute angular sensor: 12bit Hall effect based Specification Dimensions (ᴓ x L): 64mm x 110mm Weight: 525g Repeatable torque: 28Nm Maximum rotational speed: 0,58Hz Hollow shaft with 8mm diameter for cabling Different support points on drive side Multi-turn of each joint: ±360 = 720 Integrated electronics including Power electronics FPGA -based control LVDS communication 9

10 Compliance Control in General In Cartesian Space: In Joint Space:? Springer Handbook of Robotics 10

11 Torque Estimation 11

12 Identification of Torque Constant Setup2: Motor Test Bench Setup1: Winch Test 12

13 Estimation of Contact Moment by Current 13

14 Compliance Controller in Joint Space 14

15 Compliance Control using Indirect Method In Cartesian Space: In Joint Space: 15

16 Experiment Video 16

17 Current Estimation for Thorax 17

18 Experiment Result for Thorax 18

19 Experiments on Distal, Basal and Thorax 19

20 Conclusion Immediate Reactive Reflex based on Self-Awareness! With distributed compliance controllers in joint space, the legged robotic system can react faster to unexpected obstacle; can improve locomotion on uneven surfaces; reduce complexity and integration effort using estimation Thus they contribute to the improvements in overall performance: stability, energy-efficiency, better protection of the robotic system level of autonomy and robustness 20

21 Outlook and Future Work Optimization of compliance settings in various scenarios Design of a general control architecture merging reflexes, locomotion control and decision-makings Improvement in the estimation accuracy 21

22 Thank You! 22