DR. AYORKOR KORSAH
21ST DECEMBER, 2012
FINAL PROJECT REPORT
The objective of my final project was to probe further into the subject of robot communication and coordination. I wanted to gain a better understanding of how multiple robots are able to share information and collaborate on tasks in real time. Initially, I planned to design a system of robots with different capabilities that could determine how to collaborate and perform given tasks. However, I settled on implementing a system of similarly-designed robots that will collaborate on a specific task. For my final demonstration, I designed two robots that would simultaneously carry a box along a path. One of the robots would act as the master and will send driving instructions to the second robot. The second robot, the slave, would simply follow the instructions of the master.
After carefully thinking through the project, I divided it into three sub tasks – the robot design, communication mechanism and carrying of the payload.
I began by researching on NXT robot designs that could perform the task of carrying the box. After several considerations I decided to build forklift robots so that they could carry the payload off the ground. I did not readily find a design that had the forklifts on the side so I designed one myself. The robot employed a two-wheel differential drive system with a motor to power each wheel and a castor wheel at the back. It also had a third motor that powered the forklift. The design of the forklifts resulted in one side of the robot being heavier than the other which caused some irregular driving behaviour. However, I fixed this by determining the ratio of imbalance and configured the nSynchedMotors function accordingly. I carried out a number of tests to ensure that the forklifts would be strong enough to carry the box.
The next objective was to create a reliable communication mechanism between the robots. I tested out the Bluetooth functions available for NXT and realised that a lot of them were not applicable for my purposes. I ran into several difficulties at this stage of the project and had to spend a considerable amount of time resolving them. The major challenges were maintaining the connection between the robots as the robots kept losing connection on several occasions and also problems with delays in message sending. I overcame these challenges by repeating some activities to ensure increase the chance of success and also increased the amount of time scheduled for some Bluetooth-related processes. Also, by running repeated tests I managed to figure out the average time it took activities such as Bluetooth search and message sending to occur and used that as the wait time for those processes. The results of these were greater reliability and a stronger communication framework between the two robots.
CARRYING THE BOX
The challenge in accomplishing this task was to develop an algorithm that would calculate the required motor power and duration needed by each of the four motors. This drew from principles of inverse kinematics where I knew the desired robot configuration and so had to determine the configuration of the individual wheels. I carried out a number of tests to determine the motor requirements for the robots individually and recorded the results. I used this data to determine the relation between the motor powers and durations of the two robots. I used this approach in developing the synchronized movement algorithm because working with angles did not produce desired results. This was due to the fact that the motor encoders took their recordings in degrees of rotation rather than provide angular values. I tested the function several times as well and fine-tuned the time requirements to achieve near perfect synchronisation.
Putting everything together for the first time gave me heartbreak as the entire system broke down. Connection and message-sending began to fail and in effect the robots were unable to even lift the box together. A lot of it was due to the fact that it was the first time I had put everything together and had not taken a lot of factors into consideration. I had not provided any fall-backs in case communication delays were experienced and I had not accounted for natural discords in motor coordination. However, with a good number of tests and redesign, I managed to get the system to function together correctly as planned.
In doing this project I have learnt a lot about robot coordination and communication. I have seen first-hand the problems that face this field of research chief among which are communication failures. However, I managed to have a successful implementation which shows the immense opportunities that exist in the real world for such systems. It is possible to design systems that will have planet exploration robots or rescue robots share resources and join forces to carry objects or accident victims so that no particular robot is overloaded.