The fourth and final week of the FutureLearn Beginning Robotics MOOC has now come. This week the emphasis is much more on control strategies for the mobile robot and on self-learning. The same types of mobile robots are discussed but with slightly different characteristics. So there are light seekers and light avoiders and those inhibited by light. These are classed as Braitenburg robots. I had not heard of these robots before and this section was very interesting as they use a form of artificial neural networks to implement their control. However, the details for this are not included directly in the MOOC but there are links that may do so.
By small changes these can be made to operate as prey and predator mobile robots, although there is no actual eating/destruction of prey mobile robots taking place. I was very interested in this as it looked possible to move this behaviour from the simulator to real mobile robots.
There is considerable discussion of artificial life based on these simple concepts. One interesting simulation shows how birds (called boids) can swarm about just using three simple control rules.
The web site for this video is below:
Tere is then an interesting section on artificial live, for mobile robots. All good and interesting.
Being a free MOOC it is not unexpected that there is a section on the robotics courses at the University of Reading but the examples are still interesting.
The example simulations build on the concepts discussed earlier in this week and on material from previous weeks, to create a mobile robot capable of following a wall and then hence, a simple maze solving robot.
The web site for this simulation is shown below.
https://www.futurelearn.com/courses/begin-robotics/8/exercises/456475
Once you have identified a suitable set of characteristics for your simulation wall following mobile robot you can then change the environment to a simple maze to see if your mobile robot can find a way through. It is quite difficult to work out suitable values.
The web page for this is below:
https://www.futurelearn.com/courses/begin-robotics/8/exercises/456475
Then a Braitenburg version of the mobile robot is simulation where you have to select the values for the neurons/ I managed to get mine to move but that was about it.
The web page for this is below:
https://www.futurelearn.com/courses/begin-robotics/8/exercises/456477
There was the end of week quiz, which I have to say I did resort to mostly random guessing as somehow I didn't seem to have learnt the details needed to get the questions right first time. It is a good job there is no actual marking or pass/fail for this MOOC. Being the end of this course there were some outtakes of mobile robots not behaving correctly, which were amusing but also demonstrated how difficult it can be to get robots to do what you want, reliably.
Overall I thought this was an excellent MOOC and that I learned a great deal. Despite being free it contains some good content, especially in the last couple fo weeks. The first week is a little bit basic but that is not surprising for an open access MOOC such as this. I would recommend anybody to have a go at this the next time it runs. Good fun.
Dubbie