Mandatory Performance Requirements:
M.P.1 – Classify road with an accuracy of at least 85%
M.P.2 – Segment road with an accuracy of at least 85%
M.P.3 – Classify wet asphalt with an accuracy of at least 85%
M.P.4 – Segment wet asphalt with an accuracy of at least 80%
M.P.5 – Classify puddles with an accuracy of at least 85%
M.P.6 – Segment puddles with an accuracy of at least 85%
M.P.7 – Have friction prediction built into constructed environment map
M.P.8 – Planning horizon greater than stopping distance (8m at 30kph)
M.P.9 – Have real-time motion control
M.P.10 – Have online prediction model improvement
M.P.11 – Stays within 0.75 meter of the line connecting the input waypoints at all times
Mandatory Non-Functional Requirements:
M.N.1 – Be able to reach speed that is scaled equivalent to typical highway passenger vehicle speeds (30 km/hr)
M.N.2 – Maintain less than 1.47 lateral grms at all times
M.N.3 – Have representative suspension ride frequency and damping characteristics compared to passenger vehicles
M.N.4 – Use self-contained energy source
M.N.5 – Stream video and classification feed
M.N.6 – Have IP64 or better Ingress Protection rating
Desirable Performance Requirements:
D.P.1 – Classify wet leaves with an accuracy of at least 80%
D.P.2 – Segment wet leaves with an accuracy of at least 80%
D.P.3 – Classify potholes with an accuracy of at least 80%
D.P.4 – Segment potholes with an accuracy of at least 80%
D.P.5 – Classify sand with an accuracy of at least 80%
D.P.6 – Segment sand with an accuracy of at least 80%
Desirable Non-Functional Requirements:
D.N.1 – Be able to function under adverse weather conditions
D.N.2 – Cost less than the average cost of a tractor-trailer incident
D.N.3 – Generate terrain map for the user
D.N.4 – Provide warnings of incoming adverse terrain to the user
D.N.5 – Use reinforcement learning to improve vehicle dynamics