Human-robot skill transmission for mobile robot via learning by demonstration

Area of Science:

• Robotics
• Artificial Intelligence
• Machine Learning

Background:

• Mobile robots require advanced capabilities for tasks like material transport.
• Human-like autonomous target tracking is a key challenge in robot skill acquisition.
• Learning by demonstration offers a promising approach for robot skill transmission.

Purpose of the Study:

• To propose a skill transmission technique for mobile robots using learning by demonstration.
• To enable robots to autonomously track targets after learning human-like capabilities.
• To develop a framework for accurate trajectory tracking and control in real-world scenarios.

Main Methods:

• Utilized Kinect sensor for human activity recognition and planned path creation.
• Implemented dynamic movement primitives (DMP) to represent teaching data.
• Employed Gaussian mixture regression (GMR) for encoding learning trajectories.
• Investigated model predictive tracking control (MPTC) with recurrent neural networks (RNN) for accurate position control and uncertainty elimination.

Main Results:

• Successfully demonstrated a skill transmission framework for mobile robots.
• Achieved human-like autonomous target tracking capabilities.
• Validated the effectiveness of the developed techniques through experimental tasks on a BIT-6NAZA mobile robot.

Conclusions:

• The proposed skill transmission technique effectively enables mobile robots to learn and perform autonomous target tracking.
• The integration of DMP, GMR, and MPTC with RNN provides a robust solution for accurate trajectory control.
• The experimental results confirm the practical applicability and effectiveness of the developed methods in real-world robotic applications.