Community Mode: 21st Century Accident Reporting to Share Responsibility Between Industry and the Public

When a blizzard strikes in Boston, Air Traffic Control coordinates safety procedures on behalf of the airline industry. This coordination mechanism during an outlier event benefits all involved. We explore a similar if more lightweight system for AVs. Our model builds on the notion of a Shared Semantic Space, which is the area of overlap in the virtual environments that AVs use to judge the world around them. When rendered visually, this Semantic Space is often populated with purple, red, and green squares that symbolize, for an AV: cars, cyclists, pedestrians etc. To improve safety, we propose to provide vehicles access to a new common layer of Semantic Space populated by features identified by communities as risks in their area. To access this layer of contextual data, an AV would enter "Community Mode" (a play on Tesla's "Ludicruous Mode"). In Boise, for example, community advocates might flag a highway area in which homeless people are known to walk at night. This knowledge would not have otherwise been made available to an AV. In "Community Mode," AVs could access this geolocated data as well as other features that figure universally in larger contexts (ex. in America, a skateboard on the street often means that a child at play is nearby). "Community Mode" would create jobs, improve AV safety, and act as a catch-all for relevant findings generated by Go-Teams.

While the public at large could contribute to "Community Mode", it would also be populated by experts. This begs the question: when an AV is involved in an accident, who should arrive on scene to assess what went wrong and where should their findings go? Neither sheriffs nor police in the U.S. are currently trained to understand the technical systems at use in AV, which limits the extent to which they can meaningfully contribute to the improvement of these technologies. In California, they currently report using paper records. To ensure a high level of safety, we recommend that policymakers borrow from the U.S. airline industry's successful accident reporting system, which recorded zero fatalities in 2017 amongst major commercial airlines. The National Transportation Safety Board (NTSB), in partnership with the FAA, currently uses a 'Go-Team' model that employs a rotating task-force of 8-12 experts trained in technical, policy, media, forensic, and community management matters. The multi-faceted quantitative and qualitative data that this task-force generates is fed back to industry stakeholders to improve airline safety and to inform government stakeholders so they can improve the relevant infrastructure, such as Air Traffic Control systems.

Without a coordination mechanism of this type, the public would effectively have to carry the burden of risk that accompanies errors that manufacturers unknowingly share. This is an unsustainable model that may hamper the adoption of AVs by multiplying the total number of accidents by the number of industry players involved. A lightweight, coordinated system like the one we propose would improve outcomes while protecting IP by containing it within a small group of experts. This group of specialists would ensure that the right information reaches the right stakeholders.

Community Sense Maps

People working for the local City Council and conscious citizens have the latest information on what is happening in a city and where are the areas where people might have certain behaviors. This information will help inform the companies providing Autonomous Vehicles in the same areas. They can take better decisions services and in a city by providing:

  • Realtime alerts about potential hazards could be sent to multiple stakeholders - civil society organizations, the local manicipality, the manufacturing company, teams within the National Transportation Safety Board.
  • Highly annotationed maps of specific risk areas. Each area might require different parameters for the AI systems used for controlling the driving behavior of the vehicle as moves between differe neighborhoods.
For instance, on the left you see a risk area marked on the map. Every resident of San Francisco will tell you that 6th and 7th street between the Caltrain and Market st. are places where there’s a large number of homeless people. The San Francisco Library near Market street and the food services nearby as well as 101 create an environment that has been friendly to homeless people ever since the 70s. When driving through there you’d need to be extra careful and expecting people to jump unexpectedly in front of your car at any point.