Speaker: Johan Fabry
Abstract:
When thinking about robotics, the first thing that comes to mind is impressive hardware. The main problems to solve here are difficulties in building it and managing its low level functionalities, e.g. making a robot that can walk. But the robot only starts to do interesting things when you build on top of these functionalities, e.g. make it walk to a door and open it. This is almost purely a software challenge and sadly roboticists are stuck in the nineties when it comes to software development. They have an incredibly static way of programming robot behaviors. In our research we are pushing towards the total oposite: live programming of a robot, while it is running.
In this talk I will present our language, called Live Robot Programming (LRP), built on Pharo and making heavy use of Roassal for visualizations. I will show how LRP allows for the creation and evolution of robotic behaviors on-the-fly and how you can use it to play with robots. I will demonstrate LRP with a simple robot behavior, and on a real robot if I can get it to board the plane with me.
Video1: https://www.youtube.com/watch?v=0CYMkxV919s
Video2: https://www.youtube.com/watch?v=TPQXeE9yJPI
13. Waste less time in
incidental complexity
13
Use time on
fundamental complexity
14. “But why is the robot executing this behavior now?”
Example
14
“What would happen if I change epsilon to 5 ?”
(What is the internal state of the algorithm)
(What are the correct parameters for the algorithm)
15. Spend brainpower on the
complexity of the task
15
Have an immediate
connection to the
behavior
