James Bruton's Really Useful Robot-in-Progress

James Bruton has been making robots since around 2004, and started producing regular YouTube videos about them in 2013. Up until recently, his robotic experimentation has been microcontroller-based, meaning that while these robots can react very well to quick and simple stimuli, they’re not generally suitable for more complex tasks, such as computer vision.

As outlined in the video below, Bruton has now been experimenting with vision-based robotics and learning ROS, the Robot Operating System. In the clip he shows us his progress on the “Really Useful Robot,” which can drive house and uses inverse kinematics to adjust its height and arm position.

The robot is using an NVIDIA Jetson Xavier for vision processing along with an Intel RealSense camera unit for image acquisition. In addition to visual information, the RealSense device measures distances to an object, and with the proper code, he’s able to link the image recognition and depth perception information together. He’s also employing a Dynamixel servo to move the camera up and down to keep images vertically centered.

Remote control is via a Raspberry Pi and a microcontroller-based unit that looks like a great addition to the overall robot setup. While there’s more work to be done – including an end-effector for the robot arm – the robot looks very fascinating. Perhaps it will one day be able to accomplish useful tasks, such as cleaning up after Bruton when he leaves cups around the house!