Medical robotics treatment brings many benefits in rehabilitation and therapy. 3D depth cameras allow robots to visually sense their surroundings and thus adapt therapies in real time. The processing of point clouds captured by depth cameras is computationally intensive, so good optimisation of algorithms is important. As part of the thesis, we set up a measurement site, analysed the performance of the software code and the methods for processing the point cloud. We tried to optimise the processing of the point cloud by parameterisation. Improvements were tested and evaluated on an ongoing basis. The results showed that the nearest point and point cloud thinning algorithms are the most computationally demanding.
|
