Doctoral Research
Learning mobility maps from past observations of vehicle slip
My doctorial studies were aimed at generating mobility maps which were used as cost maps into path planning. The aim was to associate exteroceptive parameters (slopes, colour, texture) with speed limits. The speed limits were learned using proprioceptive feedback (slips).
Mobility Maps
Experimental results were focussed on interpreting slopes
The first goal, was to demonstrate orientation sensitive path planning in slopes (using mobility maps as an inverse cost map)
The second goal was to capture controller imperfections as velocity constraints into the mobility mapping.
S. Karumanchi, T. Allen, T. Bailey and S. Scheding, “Non-parametric Learning to Aid Path Planning Over Slopes”. The International Journal of Robotics Research (IJRR), Vol 29(8), Pages 997-1018, 2010.[pdf]