Improved Mobility Efficiency of Hexapod Robot Based on Contact Parameter Identification Using Foot-Force Sensor

Abstract

The contact parameter is crucial for the legged robot to achieve efficient walking in the field and planets. Contact parameter identification utilizing foot force sensors is a good solution, especially when the vision sensor cannot feature the scene. In this article, a novel force sensor on the foot of legged robots to identify the contact parameter and classify the ground is proposed. The proposed lightweight foot-force sensor using pressure elements is integrated with the flexible hemisphere foot made of hardened plastic-like material. Based on the artificial neural network, the contact force is obtained by the array measurement of those pressure elements while the legged robot is walking. An online method of contact parameter identification based on the multilayer perceptron is proposed and an online classification method utilizing pretrained neural networks based on ResNet is built to guide the trajectory planning of the hexapod robot. An adaptive gait of hexapod robots utilizing the swing sunken height fitting contact parameters is designed to improve mobility efficiency. The proposed contact parameter identification and the ground classification are verified on our hexapod robot using the proposed foot-force sensor. Using the proposed method, the hexapod robot can exchange the swing trajectory based on the contact parameters of different terrains to reduce mobile energy consumption.