Roboceratops by Robert Stephenson [Archive] - Forums - Portal of Robotics and Artificial Intelligence

Airicist

12th May 2017, 20:55

https://youtu.be/uPbS0CFx-PY

Roboceratops - Up and walking

Published on May 12, 2017

This is Roboceratops, my Robot Dinosaur.

Inspired by a childhood love of dinosaurs, I wanted to try and make a member of the ceratopsian group. And to attempt to give the feel and personality of a juvenile dinosaur.

Finally, after nearly three years, I had two weeks of downtime at work, so it was time to dust off the dino, make some changes and test some theories.

Changes:-

Mechanics:-
The number of degrees of freedom was the limiting factor with Mk1. The front legs being the major culprit. This was rectified on Mk2 by adding the third degree of freedom to them. This was tricky while still trying to keep the form factor as small as possible.
The Robot now has 16 Degrees of Freedom using the following;

Jaw - 1x 9g Servo
Neck - 2x Hitec HS 55
Tail - 1x Hitec HS 645 MG
Front Legs - 2x Hitec HS 645 MG & 1x New Power XLD - 45 HMB ea
Rear Legs - 3x New Power XLD - 45 HMB ea

Certain servos had to be upgraded due to torque requirements, but as I am working on a budget rather than upgrading within the hitec brand I had to swap to New Power XLD - 45 HMB's with 15kgcm torque compared to the 9kgcm of the HS 645 MH.

The head and body are made from laser cut MDF, 3 and 6mm respectively. The legs were changed, from aluminium square bar covered with upholstery foam, to layered laser cut 6mm MDF.
Then the whole structure is sculpted by sanding to give organic, bone like, shapes and forms.

Code:-

I have revised the code to more effectively utilise the extra degrees of freedom.

The new Transformation Matrix based Forward Kinematics allow the robot to translate the body in each axis, fore-aft, left-right and up-down. Also rotate in pitch roll and yaw, with the centre point of the rotation being movable, eg: at the centre of the robot, back of the head or on the floor etc.

The revised Inverse Kinematics takes into account the extra degree of freedom in the front legs which allows each of the four legs use the same IK function by passing parameters to the function rather than having custom IK functions for each leg.

The Gait Engine has been rewritten using the basis of a Hexapod Gait Engine. Extremely parameterised, the engine has not yet been fully tuned for the different requirements of a quadruped with legs underneath versus a hexapod with legs out to the side.
The main thing with quadruped locomotion is the centre of gravity shifting required to maintain stability on less than four legs.
I have parameters for step height, step speed, step duration, amount of C.O.G. shifting. These will all need tuning slightly to get the best out of the robot.

Now Mk2 has acted as a proof of concept that this platform is able to locomote, the next stage over the next coming years, when time and money allow, will be to scale from standing 1ft tall up to possibly 3ft tall.

One day take my robot dino for a walk on a leash... One day...