- Physical robots can be inconvenient
- Especially, they tend to be both expensive and fragile
- This is really obvious with rockets, but it's also true for less dangerous
robots like the Ranger.
- Again, the solution is simulation.
- To simulate wheeled robots, we'll be using Gazebo
- Gazebo is an open source robot simulator that's commonly used in academia and
sometimes used in industry.
- It's normally used with the ROS software stack, but we're just going to use
Demo and HW01
- In HW01, you'll be setting up your computer to run Gazebo and then writing a
simple control program for a simulated robot.
- Gazebo supports running on Linux. Officially, they require a recent release
- I'm running Debian Linux (very similar to Ubuntu) to develop the assignments.
- The two supported configurations will be Debian 10 or Ubuntu 20.04 directly
installed (e.g. dual boot) on your personal computer.
- If you can get it to work some other way, that's fine, but if it doesn't the
course staff can't help you with your unsupported setup.
- Gazebo does OpenGL rendering, which is likely to break several common
alternate setups (e.g. running in a normal VM).
Show the solution code running.
- Once you have Linux and Gazebo installed, you're ready for HW01.
- Download the starter code from Inkfish.
- Run Gazebo with
- Goal: Write a control program that drives the robot car from start to checkerboard.
- Show the brain directory.
- "Write a control program".
- Run the solution brain program.
- Sensors: Laser scanner, pose sensor.
- Actuators: Turn wheels, steering.
Submitting the Homework
- Create a "brain-hw01" directory.
- Edit the README, add a video (the HW01 solution video)
- Create a tarball
- Submit on Inkfish