Browsing by Author "Van Zyl, Rynhardt"
Now showing 1 - 1 of 1
Results Per Page
Sort Options
- ItemLow-cost autonomous rover with heterogeneous compute(Stellenbosch : Stellenbosch University, 2023-12) Van Zyl, Rynhardt; Fisher, Callen; Stellenbosch University. Faculty of Engineering. Dept. of Electrical and Electronic Engineering.ENGLISH ABSTRACT: This research investigates the development of a cost-effective autonomous rover system, utilising a heterogeneous computing architecture to meet its operational goals. The central question addressed is: Can a budget-friendly autonomous rover system be constructed that effectively utilises a heterogeneous computing architecture to meet its operational objectives? The study begins with a detailed review of current advancements in low-cost autonomous rovers and heterogeneous computing architectures. This review identifies existing challenges and opportunities in this field, providing a solid foundation for developing the autonomous rover system. A low-cost autonomous rover system incorporating a heterogeneous computing architecture was designed and constructed. This system, priced under R50 000, operates with minimal human intervention, receiving only goal-oriented instructions from an operator. The embedded heterogeneous computing architecture utilises multiple tiers of computational resources, distributing workloads across various processing units, thereby avoiding exclusive reliance on multi-threaded computation. The performance of the developed system was evaluated in a controlled, real-world environment. The system demonstrated high adaptability and efficiency in different conditions, with an average speed of 0.5 m/s and an average energy consumption of 0.2 kWh (measured by an external device). The system also demonstrated high adaptability, with an error rate of less than 5% in decision-making. The average decision-making time was 0.1098 seconds, with an accuracy rate of over 95%. The system was found to have a high task allocation efficiency of over 90%, with an average execution time of less than 0.1 seconds. The findings of this research have the potential to significantly contribute to the field of autonomous systems, providing a blueprint for the development of cost-effective solutions for autonomous navigation and operation. The study concludes with a discussion of the implications of the research and provides a trajectory for future endeavours in developing budget-friendly autonomous rover systems harnessing the power of heterogeneous computing architectures.