A simple mechanical robot that can avoid cliffs using a castor and elevated fly-wheel that evokes right-turns as soon as the castor crosses the edge.

The ASPIRE Engineering Summer Bridge program is an offering of CU’s BOLD Center to incoming freshman to get ready for their freshman year by taking introductory classes and to become acclimated to Campus life.

We host four cohorts of ASPIRE students for a 30-min “crash course” in robotics today and tomorrow. We first show them a mechanical robot that can avoid cliffs using a fly-wheel mechanism as an example how important mechanical design can be to create the (illusion?) of intelligent behavior. We then let them brainstorm on how they would program  a robot to solve the “Ratslife” contest. As the students usually do not know what capabilities a robot possibly can and cannot have, we ask them questions like “What would you do when put in a maze and have to find a water fountain?”. Here, students immediately come up with very sophisticated algorithms that make heavily use of their own cognitive and sensorial capabilities. We can then ask “Ok, and what about if you were blind?” or “What if you don’t have pen and pencil to draw a map?”, which let them come up with the simple algorithm of performing wall-following combined with doing strict right turns. Interestingly, this algorithm could be implemented by a purely mechanical device using a similar mechanism as the cliff-avoiding robot, not requiring any fancy hardware.

The "Ratslife" environment with a 2011 ASPIRE cohort.

We then get back to their first plan of using cues in the environment to come up with a map and do systematic exploration. We explain then that a robot such as the e-Puck robot that has a little camera, could recognize individual markers and use its “odometer” to relate them to each other to get a map. This map could then be used together with the wall-following algorithm that the students came up with earlier (technically resulting in a Depth-First-Search) to map chargers and then find shortest paths between them.

At this time, some hints can be given to the SLAM problem. We can ask the students, “You see the Eiffel-Tower, where do you think you are?” (Paris), and then “Where else could you be?” (Las Vegas). “Assume you know you are pretty certain that you are in Paris  and take a 10h flight towards west, leaving the plane you see again an Eiffel-Tower, where are you now?” (Most likely in Vegas, except you flew in a circle!).

Simple solutions to the shortest-path problem (such as Dijkstra’s algorithm) can be very intuitively explained on the Ratslife arena. We ask the students to imagine a rubber-sheet being spanned over the arena. We then ask them to imagine the rubber sheet is lifted up where the robot is and pressed down at a charger’s location. What happens if you would put a marble on the highest point on the sheet? Well, it would find its way to the deepest point, following the shortest path! This is pretty much what Dijkstra’s algorithm does, just using numbers on a grid and then following the steepest gradient.


Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Set your Twitter account name in your settings to use the TwitterBar Section.