Robot trajectory learning through practice

We present an algorithm that uses trajectory following errors to improve a feedforward command to a robot. This approach to robot learning is based on explicit modeling of the robot; and uses an inverse of the robot model as part of a learning operator which processes the trajectory errors. Results are presented from a successful implementation of this procedure on the MIT Serial Link Direct Drive Arm. The major point of this paper is that more accurate robot models improve trajectory learning performance, and learning algorithms do not reduce the need for good models in robot control.

Authors

• Christopher G. Atkeson
• Joe McIntyre

Keywords

• Adaptive control
• Error correction
• Actuators
• Robot sensing systems
• Learning
• Feedback
• Convergence
• Performance analysis
• Algorithm design and analysis
• Artificial intelligence
• Learning Trajectories
• Learning Algorithms
• Learning Performance
• Robot Control
• Robot Model
• Trajectory Error
• Robot Learning
• Simple Model
• Use Of Models
• Error Model
• Feedback Control
• Moment Of Inertia
• Robotic Arm
• Output Control
• Perfect Model
• Sensor Noise
• Finite Interval
• Velocity Error
• Joint Velocity
• Unmodeled Dynamics
• Joint Acceleration
• Rigid Body Dynamics
• Non-minimum Phase
• Robot Dynamics
• Simple Dynamic Model
• Robot Joint
• Final Trajectory
• Dynamic Model
• Rigid Model