DIPLOID -The Quadruped Robot

DIPLOID is four-legged Quadruped robot inspired by the Boston Dynamics Spot and MIT Cheetah, being developed by Team Robocon BITS Pilani.

Started in early 2019 as a long-term project with an aim to achieve stable walking on rough terrain and environment perception using SLAM technique. Jump to:

• Motivation
• Overview
• Kinematics
• Gait
• Simulation
• Chassis and Hardware
• Future Goals
• References
• Special Thanks

Motivation

Quadruped robots have several applications like last mile delivery, guiding robots for blind, security and surveillance, accessing hazardous enviornments, space exploration, etc.

The Quadruped robot community and research is growing everyday in different parts of the world. So far, there hasn't been a robust quadruped on an undergraduate level in India, and we on the journey of develpoing one.

Overview

This project is the source code for DIPLOID quadruped, a 4 legged robot. The software is implemented on a Raspberry Pi. As of now we have achieved-

• Gaits like- trot, walk, pace, gallop and turning
• Gazebo simulation for testing gaits and RL

Jan 2021 update:

• ros publish code hasn't been added
• This repo is on hault. New repo present on bitsrobocon github is the one being updated

Kinematics

The inverse kinematics has been demonstrated in the illustration below:

Source: Spotmicroai community

Gait

This is how various walking gaits look like in a simple 2D animation.

Various gaits implemented so far.

Simulation

A simulation with trot walking gait

A trial simulation with pace gait

Chassis and Hardware

The design for legs is inspired from spot-micro robot CAD but has been modified to incorporate Dynamixel AX-12A servo actuators.

Hardware planned to be used:

• Raspberry Pi 4
• Jetson nano
• Intel Realsense
• Dynamixel AX-12A actuators
• IMU

Future Goals

We are aiming for stable walking on uneven terrains, blind stair climbing capability and developing an autonomous motion planning feature of the robot to execute a simple task around a sensed 3D environment while avoiding obstacles or possiblities of imbalance on its own.

References

• Original idea : MIT Cheetah and Spot by Boston Dynamics
• High speed trot-running: Implementation of a hierarchical controller using proprioceptive impedance control on the MIT Cheetah
• https://github.com/mike4192/spotMicro
• Dillinger, open source with a public repository on GitHub.

Special Thanks

• Spot mini mini : authors - Maurice Rahme and Ian Abraham and Matthew Elwin and Todd Murphey spotminimini2020github
• Spot Micro AI Community : https://spotmicroai.readthedocs.io/en/latest/