Biologically-inspired adaptive obstacle negotiation behavior of hexapod robots

Area of Science:

• Robotics
• Neuroscience
• Biomimicry

Background:

• Insects exhibit adaptable leg movements for obstacle negotiation.
• The distance to an obstacle is critical for successful insect climbing.
• Hexapod robots require adaptive control for complex terrain.

Purpose of the Study:

• To develop an adaptive neural control mechanism for hexapod robot obstacle negotiation.
• To enable robots to autonomously adjust climbing initiation based on environmental cues.
• To integrate locomotion, reflexes, and learning for adaptive behavior.

Main Methods:

• Implemented a control system combining locomotion, joint control, leg reflexes, and neural learning.
• Utilized ultrasonic sensors to provide conditioned (CS) and unconditioned (UCS) stimuli.
• Trained the neural learning mechanism to associate predictive and reflex signals.
• Tested the system in physical robot simulation and on a real hexapod robot (AMOS II).

Main Results:

• The robot learned to autonomously determine the appropriate distance to initiate climbing.
• Successfully negotiated obstacles up to 85% of leg length in simulation.
• Achieved successful negotiation of obstacles up to 75% of leg length on a real robot.

Conclusions:

• The adaptive neural control mechanism enables efficient obstacle negotiation in hexapod robots.
• Bio-inspired learning allows robots to adapt to variable obstacle heights and gaits.
• This approach advances autonomous robotic locomotion in unstructured environments.