Bachelor-Thesis
Designing a High-Speed Actuated Table for Balancing Miniature Poles
Pole Balancing, e.g. keeping a long stick or broom upright on your fingertip, is amongst the standard evaluation tasks for robot controllers. We have recently developed a balancing robot (hardware and software) that can balance small poles, such as ordinary pencils or chop-sticks, using visual feedback only. Balancing such small objects is not possible with traditional cameras; instead we use an event-based vision sensor that emits up to 200.000 individual messages ("events") per second whenever the pencil moves. The visual input - we believe - would allow much smaller objects to get balanced, such as toothpicks or matchsticks. The current robotic hardware, however, is by far too slow and too imprecise for objects smaller than pencils.
This current hardware setup consists of two independent servo motors that each actuates a small table in orthogonal direction (see image below). As the internal controller of those servo motors is not tuned to our hardware, these motors introduce significant delays in actuation and often overshoot the desired target position, which severely limits balancing performance.
Hence we are interested in an alternate option for actuation. The student will investigate in magnetic actuation of a lightweight table:
- Explore possible options for actuation (e.g. linear magnetic instead of motor/servo)
- Characterize requirements for such a system (compute resulting forces for given mechanics, power, etc)
- Build such a system (this will be a major part of the thesis!)
- Program a controller for such a system in C on a microcontroller
- Provide an interface for other controllers / hardware
If successful, we envision the system to be able to balance small rods such as toothpicks or matchsticks, which no-one has ever done before! This will be an impressive demonstrator!
Student:
Francisco Javier Llobet Blandino
Advisor:
