Parallel algorithms for target tracking on multi-core platform with mobile LEGO robots

The aim of this master thesis was to develop a versatile and reliable experimental platform of mobile robots, solving tracking problems, for education and research. Evaluation of parallel bearings-only tracking and control algorithms on a multi-core architecture has been performed. The platform was implemented as a mobile wireless sensor network using multiple mobile robots, each using a mounted camera for data acquisition. Data processing was performed on the mobile robots and on a server, which also played the role of network communication hub. A major focus was to implement this platform in a exible manner to allow for education and future research in the elds of signal processing, wireless sensor networks and automatic control. The implemented platform was intended to act as a bridge between the ideal world of simulation and the non-ideal real world of full scale prototypes. The implemented algorithms did estimation of the positions of the robots, estimation of a non-cooperating target's position and regulating the positions of the robots. The tracking algorithms implemented were the Gaussian particle lter, the globally distributed particle lter and the locally distributed particle lter. The regulator tried to move the robots to give the highest possible sensor information under given constraints. The regulators implemented used model predictive control algorithms. Code for communicating with lters in external processes were implemented together with tools for data extraction and statistical analysis. Both implementation details and evaluation of di erent tracking algorithms are presented. Some algorithms have been tested as examples of the platforms capabilities, among them scalability and accuracy of some particle ltering techniques. The lters performed with su cient accuracy and showed a close to linear speedup using up to 12 processor cores. Performance of parallel particle ltering with constraints on network bandwidth was also studied, measuring breakpoints on lter communication to avoid weight starvation. Quality of the sensor readings, network latency and hardware performance are discussed. Experiments showed that the platform was a viable alternative for data acquisition in algorithm development and for benchmarking to multi-core architecture. The platform was shown to be exible enough to be used a framework for future algorithm development and education in automatic control.