Storms often provide a test for British infrastructure and the early months of 2022 saw Storm Eunice, swiftly followed by Franklin, give the sternest of examinations.
Sadly the UK transport infrastructure weakness in such an environment was exposed with more pressure being put on the UK roads as commuters would turn to cars to get by. Airports saw diverted and cancelled flights with the UK rail infrastructure running skeleton services as fallen trees scattered the tracks.
Driving into London from my home in Wiltshire was a challenging experience during Storm Franklin. Debris from trees scattered the roads with floods in areas creating diversions to make the journey longer.
Humans are able to assess their path and then consider the challenges on the road to make a rational decision on the course of action ahead as I did on this journey. A question that occurs to me is how an Autonomous Vehicle might act in such an environment.
People have subjective factors that keep us and others safe, one of them being self preservation. A slightly flooded country road will mean I can judge if water is shallow enough for me to drive through. An AV Car would have the same dilemma but how would it act?
Would the AV Car calculate the risk and judge the deepness of the flooded part of the road and thus give due safety to passengers and those around the vehicle? What is an easy situation for ourselves is possibly not the same for an AV. The ease we have in terms of judging how deep a flooded road is, might be too complex for an AV.
Another scenario that a storm could bring is obstacles from buildings falling into the road. In such a case we as humans would never try to just drive through such an obstacle, would an AV?
Driving on the M25 large lorries would be driving at 30 mph in an effort to avoid being blown around and in some cases over by the blowing storm. Potential dangers for the small vehicles driving at a normal speed on the motorway. Another scenario for an AV Car. A scenario that developers and makers of AV cars would have to consider.
Engineers of Autonomous Vehicles will be looking at different scenarios all the time, hundreds of different scenarios. Getting the correct eventuality of these scenarios correct is the big challenge. Climate change will bring predictably unpredictable weather to the UK roads over the next few years, predicting and calculating these scenarios is one of the many challenges for AVs.
∂RISK works every day to prepare AVs for the scenarios and complex situations on the road. We focus on edge-cases, the one in a million high risk scenarios and situations that AVs should be prepared for.
A person at a traffic light in a green coat about to cross the road on a red light and the pedestrian walk sign is an easy spot for a human but an AV might see the green coat as GO. This is an example of a high risk situation and what is meant by an edge case.
∂RISK uses real-life data from road and traffic cameras, accident reports Insurance claims, and peoples experiences. We constantly gather information to create the world’s largest library of edge case scenarios.
∂RISK then takes the world of data to create simulations for training AVs for each and every scenario. Information and training from our Edge Cases allows AVs to see these hazards and scenarios before they even occur.
For AVs to not just be regulated properly but to have confidence with the general public, the must be tested for all possible scenarios and situations for the road.
The testing capabilities that we have developed at ∂RISK means AVs are already being tested and trained on edge cases of extremely rare events that can happen. Testing, training and validating on ∂RISK edge cases will mean that AVs will be safe in storms and all the fallout they may bring.