For humanoid robots to replace humans in dangerous environments, they require a robust fall prevention system that can detect falls, prevent major damage and recover back up to continue on with the task. This paper introduces a system that is capable of fall detection, mitigation and recovery in a robot equipped with walking sticks. Walking sticks allow the humanoid robot to extend its support polygon in a fall. Using IMU data and kinematics, the orientation of the robot is used to detect if external disturbances would cause a fall. The required position of the walking stick tips are calculated to minimise the angle fallen before the system is deployed. A recovery sequence is generated to recover the robot to a stable state prior to the disturbance. Hardware experiments on the Robotis OP2 found the system was capable of reducing the impact shock for forward falls while standing and walking.
Benjamin Tam & Navinda Kottege, (2016), Fall Avoidance and Recovery for Bipedal Robots using Walking Sticks. In proceedings of the Australasian Conference on Robotics & Automation (ACRA 2016), Brisbane, QLD.
[Abstract][Paper] Download citation [BibTex]