Safe Robot Trajectory Control Using Probabilistic Movement Primitives and Control Barrier Functions.

Area of Science:

• Robotics
• Control Theory
• Machine Learning

Background:

• Probabilistic Movement Primitives (ProMPs) are effective for robot motion planning.
• Ensuring safety and minimizing control effort in ProMP-based systems remains a challenge.
• Existing methods often lack guarantees for maintaining system state within desired bounds.

Purpose of the Study:

• To develop a novel control design for ProMPs that guarantees safety.
• To ensure robot trajectories remain within a defined distance from the ProMP distribution mean.
• To allow for prioritization of safety objectives in control design.

Main Methods:

• Utilized control barrier functions and control Lyapunov functions defined by ProMP distributions.
• Employed feedback linearization to manage nonlinear system dynamics.
• Applied real-time quadratic programming for constraint satisfaction and effort minimization.

Main Results:

• Demonstrated that the robot can move along trajectories within the ProMP distribution.
• Guaranteed that the system state stays within a desired distance from the distribution mean.
• Showcased the ability to prioritize safety objectives.
• Validated the approach through simulations and real-time experiments.

Conclusions:

• The proposed control design effectively integrates safety guarantees with ProMP-based motion.
• The method offers a flexible way to manage safety constraints and control effort.
• The approach is computationally efficient, suitable for real-time robotic applications.