Gait Generation for a Humanoid Robot

Exhibitor: Tim Pike

Supervisor: Gordon Wyeth

Research Group: Complex and Intelligent Systems

Industry Sector: Scientific / Research Services

The future goal of the GuRoo project is that of one day competing in the Robocup world robot soccer competition with other humanoid robots. To this end, one of the major components of a humanoid robot is the ability of that robot to walk. To that end, this thesis looks at the development of a gait generation algorithm including the development of straight line walking using that algorithm.

The primary achievement of this thesis was the development and implementation of a gait generation algorithm that allowed GuRoo to walk in a straight line. The gait generation algorithm developed was a static algorithm that is parameterised to allow the easy changing of aspects of the walk, ie. step size etc. This allows a much more dynamic and flexible system. Also the algorithm was developed with a number of changeable variables in mind, that currently are fine tuned using a genetic algorithm, but can easily be modified to be changed dynamically using feedback from the actual robot.

Also among the achievements of this thesis, was the continued development of the simulated GuRoo environment, including the design and implementation of a new high level behaviour control system that allows simultaneous behaviours (ie. crouching and torso movement, or walking and torso movement) that are independent of each other to run concurrently on the robot (both in simulation and real life). This is a important achievement since the future work on the GuRoo software will include moving to a more sophisticated behaviour management system, which will hopefully be built on top of this implemented system.

Thesis Document (PDF)

Movie Download