Virtual Robotics - The idea

I think that there are two use cases for good robotic simulation: education and prototyping.

Education

Robotics is fun, but it can be challenging to get into. It’s not nearly as easy as picking up a python book and making cool starter programs. It would be better to separate the hardware design and the software. With a virtual robot, a student could learn how to read data from the sensors and drive the servo motors. They could also see what happens to their virtual robots running their code in different environments. Having detailed debugging tools could allow the student to see exactly why their robot got stuck in the corner and kept running into the wall. After the student got a handle on the basics, they could put their virtual robot into more realistic environments where they could experience wheel slippage and sensor data degradation.  Running in more realistic environments could allow them to slowly learn to write robust software and even make unit tests for the robot in specific challenging areas. The last piece is to allow the student to modify the hardware in their virtual robot and see how their code would work in a slightly different robot or what more fine-grain sensors and motor controllers could allow their robot to do.

Prototyping

Serious robotics companies and teams have custom software to simulate their robot behavior. Many of them also have hardware-in-the-loop simulation. The software that I’m proposing here isn’t fundamentally different than what many already have. The biggest difference is that Virtual Robotics would be a flexible simulation framework where companies could tailor it to their needs be them aerospace, ground based automatic vehicles or even those that swim underwater. This would make developing custom simulation software from scratch unnecessary.