Abstract:
Robots can be designed and deployed in water for a wide array of tasks in order to save time, money and manpower. Potential applications could be life saving, coastal surveillance & mapping, oil spill detection and mapping, fish protection from predatory birds, water bodies hygiene monitoring and maintenance, mine sweeping and degaussing, internal inspection of large tanks and mobile platforms for underwater inspection of marine structures and ships. However to qualify for such a deployment the watercraft hosting the robot should demonstrate intelligence and autonomy, either partial or complete in understanding and responding to its environment before higher level tasks are pursued. This intelligence and autonomy must be manifested in the ability of the robotic watercraft to execute autonomous navigation.
From a vision guided autonomous navigation standpoint the two major components are vision and navigation. The uniqueness of the terrain mutates these two components in comparison to autonomous navigation on land. Specifically, absence of pathways makes water based navigation easier than its land based counterpart. Simultaneously, fluid nature of water makes it difficult to prepare a valid mathematical model or extract relevant features and consequently the vision component becomes difficult to implement; both in the design as well as the deployment phase. Whilst the navigation component is not difficult to achieve, the overall process of vision guided autonomous navigation is gravely hindered on account of inability of vision component to identify the foreground objects and their related parameters.
Current research will build on existing literature from eclectic fields including but not restricted to robotics, water navigation, signal processing, image processing, textures, fingerprint analysis and pattern recognition. Unknown obstacle detection will be attempted in both the temporal as well as the spatial domains with emphasis on efficiency. Various algorithms borrowed from these two domains will be analysed. Results from pilot studies conducted on in these two domains will be presented with the research plan to develop a robust autonomous navigation system for robotic watercrafts.
