In order to have a robot to perform a task successfully, it is required to introduce the environment to the robot. This process is called teaching. Teaching process is a time consuming, expensive, and sometimes boring process for those industries that use robotic technologies.
To overcome this issue, as our undergraduate senior design (ME102 course) at University of California, Berkeley, we decided to automate this process. This means that the robot should be able to scan and recognize its environment, which is arranged randomly, and calibrate itself automatically according to its surroundings. Upon successful calibration, the robot should perform a process to demonstrate its ability for successful automatic self calibration.
The following video shows 3 stages of our robot: 1- homing, 2- self calibration, 3- process.
In this video, we moved the stands randomly and the robot was able to recognize the exact location of them and perform the desired task.
By automating teaching process, robotic industry may enjoy faster, cheaper, yet more precise calibration.
In the following video, our professor's 6 year old son, Sebastian, tried to challenge our robot by putting his robot toys standing on the blocks while our robot arm moved the blocks. Interestingly, the robot toy stood on the block during the whole time of transition without falling!
The following video focuses on 2 steps of calibration: 1- quick scan, 2- presice scan. In quick scan, robot quickly scans the environment to get a rough idea of the location of 2 stands around it. Then, by knowing each stand's approximate location, it can get closer to each stand and perform a more precise scan from closer points.
