If mechanical devices are ever going to become part of society, they must be trusted as safe, reliable and easy to interact with; controlling humanoid robots in a realistic, efficient and safe manner is vitally important.
In this session, I will present the software engineering techniques, algorithms, tools and technologies we used in the intelligence and control system for a humanoid robotic torso consisting of a head with two infrared cameras for eyes, and an arm.
This will be a whirlwind introduction to how you might go about building a robotic brain, and making sure that the brain doesn't cause the robot (or, more importantly, its operators) any harm. I'll cover the networking techniques implemented, the health and safety considerations, the numerous programming languages used for various system components, our mission to use as many different platforms, programming languages and frameworks as possible, and a brief description of what it's like to see your £2,000 robotic arm destroy itself by smashing into a table. There will also be videos of the robot in action, under automated control as well as manual control via the use of an Xbox Kinect (unfortunately the University won't let us take the robot outside to play, and for good reason).