Reference Trajectory Adaptation to Improve Human-Robot Interaction: A Database-Driven Approach.

Area of Science:

• Robotics
• Rehabilitation Engineering
• Biomechanics

Background:

• Robotic therapy and assistive devices often use fixed trajectories.
• These fixed paths can cause users to adopt unnatural movements, potentially hindering rehabilitation.
• There is a need for adaptive robotic systems that conform to individual user biomechanics.

Purpose of the Study:

• To develop a database-driven approach for adapting robotic device reference trajectories.
• To enhance robotic training for cyclic motions like walking by personalizing movement patterns.
• To improve the efficacy and user-friendliness of robotic rehabilitation devices.

Main Methods:

• Implemented a database-driven system to adapt reference trajectories for cyclic robotic motion.
• Utilized dynamic time warping (DTW) to compare user movements against a database of safe trajectories.
• Selected the best matching trajectory from the database for real-time movement assistance.

Main Results:

• The proposed adaptive trajectory method was evaluated using computer simulations and a robotic test bench.
• Demonstrated a reduction in root-mean-square (RMS) error between individual user trajectories and the reference trajectory.
• The system showed robustness to measurement noise, maintaining performance in simulated noisy conditions.

Conclusions:

• The database-driven adaptive trajectory approach offers a promising method for personalizing robotic rehabilitation.
• Dynamic time warping enables robots to adjust to individual user movement patterns effectively.
• This technology has the potential to improve rehabilitation outcomes by providing more natural and effective robotic assistance.