Slate Robotics

Dynamics! Demo of a joint computing torques from the rest of the arm’s position so that we can separate forces due to motion and gravity and outside forces from the user or the environment. This gets more complicated as the joint needs to know what the rest of the arm is doing, but as you can see, we’ve got an initial iteration of this working.

Getting closer to the desired end-state of these controllers. This is more demos of the zero-torque control and positional compliance. Lots of progress. Will have a demo of the trajectory planner pretty soon as well.

Compliance! Cascading motor controller with position, velocity, and torque control. This means that you can bump and hit the robot while it’s operating and it will comply with unexpected forces. Important for safety and other features down the road like zero-gravity mode.

It’s looking very, very smooth now. Some better control algorithms and a few minor hardware changes. This is a very heavy arm with very powerful motors on a mobile platform, so quick changes in the velocity can make it bounce. But it’s running really quite well now.

Wire covers and actuator housings installed This week, going to implement feedforward pathways from the torque measurements into the control loop in order to rely less on the I gain in the velocity controllers. This should make the movements even smoother.

Robot butlers are the future

Looking pretty dang good! Joint coordination is excellent. Need to work on computing better deceleration as we approach target positions, but I’m very happy with the progress on the controllers this week!

The controllers in each actuator now plan out velocity trajectories based on a pre-determined time that all joints should arrive at a given position. This allows for just a single command to be sent to each actuator per waypoint. A little more tuning to do, but it’s looking great (minus the wires).

Gripper testing and tuning. Going to redesign one of the links in the arm to be more rigid to keep it from bouncing. Also, the controllers need to be tuned a little bit more. All in all, it’s working well and the whole package is starting to come together.

Pick-N-Place app running on the robot. This makes it easy to create programs and get the robot to perform repetitive tasks.

TR software in the works: Pick-n-Place. Create programs quickly and easily plan out paths for the robot arm to follow. Watch as we create an entire program in under 60 seconds!

I’m not sure how Carry 300 pound man ended up in the spec sheet, but alas, here we are.

Navigation and mapping working on the real robot!

Here’s a better demo of the GUI in action. Point and click navigation. Real-time mapping. It’s like playing a video game.

SLAM implemented with working minimap navigation. Point and click where you want to go, and it maps the environment in real time.

Working on a proper teleop application for our robots. Switch between first and third person view, live minimap, control menu, etc. Wanting to incorporate SLAM next and load a map into the program on the fly allowing you to click on the minimap to navigate. Long way to go, but this gets the idea across.