for a school project, we decided to build a robotic dog. We designed and 3D-printed it ourselves. Our prior experience includes only about 4-5 months of C/C++ in Arduino (PlatformIO), so it’s rather amateurish, but it doesn’t has to be perfect so if it sort of works it is enough.
The first walking movements are functional but of course need improvement. The main issue is that the robot can’t lift a leg without tipping in that direction. We’ve shared a short video in Dropbox showing this. We’re considering moving the hind legs further forward to shift the center of gravity, but we’re unsure what other specific changes we need to make to solve this problem. The material is already quite lightweight, so we’re uncertain if reducing the body- and/or legsize would offer much benefit.
Great work. The issue you are facing is two-folds, the first one is grip of the robot, perhaps try adding some rubber pads to the foot. The second issue, which is the major one is that to avoid tipping your control is using static gait. The static gait timings have to be adjusted either based on the mechanics of your robot (i.e. model based control) or you can try training it on Isaac Lab for dynamic gait planning. The robot has to move the position of the 3 other legs to prevent the tilting.
Usually instead of the static position of the foots being 90 degrees below, its worth trying to make it 110 degrees facing outside so that in all cases with 3 legs it is always stable. You can refer to the citations at the end of the post to books so that you can create your own quick gait planner that compensate for the weight shift in the robot.
