Estimated emergency vehicle crashes in the United States cost $ 35 billion per year, and collision deaths are 4.8 times higher for emergency responders than the national average. Police officers have roughly double the rate of motor vehicle accidents per million vehicles driven than the public.
During a session at IACP 2021, panelists defined the problems and discussed potential solutions that Thomas Lu, Ph.D., and Edward Chow, Ph.D., of the Jet Propulsion Lab (JPL) are working on. from NASA to the California Institute of Technology. .
Reliable and explainable police artificial intelligence
JPL has developed artificial intelligence (AI) platforms for use in space exploration. JPL has been experimenting with the transition of this AI into the public security space for three to four years. This type of AI could improve agent safety through enriched 360-degree situational awareness.
The sheer volume of data from next-generation communication tools and sensors threatens to overwhelm or distract first responders from their critical activities. Information overload can create barriers for first responders to perform their tasks safely and effectively. This applies not only to first responders in the field, but also to those responsible for managing and leading an incident response from a higher level.
To provide solutions, JPL led a project funded by the US Department of Transportation’s National Highway Traffic Safety Administration to study technology to improve the safety of first responders and crews in and around active traffic. This resulted in the creation of the Reliable and Explainable Police Artificial Intelligence AI Assistant (TruePAL), which provides real-time warnings of risks by analyzing the environment and traffic patterns to generate a warning in time. timely to drivers and on-board crews to avoid accidents.
The TruePAL team included members of JPL from NASA, Temple University and the Miami Dade, Florida Police Department and was conducted in two phases. Stakeholders were engaged to identify key challenges and use cases and TruePAL was developed and tested in a simulated environment to demonstrate the feasibility, follow-up testing and validation of TruePAL with real traffic accident data first responders.
IT scenario test
To develop AI responses within the TruePAL system, JPL has developed computer models, using the CARLA driving simulator, of several common scenarios that can put vehicles and crews at risk. They were:
- Safety at intersections: Involve the police, EMS, fire brigade and breakdown services on the way to an accident. Modeled TruePAL helping drivers cross intersections.
- Road safety : TruePAL provides 360 degree situational awareness for vehicles parked at the roadside and warns vehicles of a potential collision.
- Identification of the danger sign: TruePAL detects the danger sign of the truck and transmits information to the relevant actors on hazardous or flammable materials.
- First aid assistant: TruePAL guides responders to prioritize first aid and perform corrective procedures.
- Electric vehicle guide: TruePAL guides the responder to correctly manage the electric vehicle and the battery involved in the accident.
The future of TruePAL
The end game of the study will be to provide an AI system that examines and manages all inputs from onboard distractions (radio, radar computers, on-board sensors, scanners, cameras, etc.) and provides situational awareness to 360 degrees and recommendations for action. All of this will be fed back to the driver via a head-up display (HUD) and voice interface. In response, the driver will be able to communicate with the on-board systems via voice commands using a chatbot (think Alexia or Siri!).
Once developed, JPL’s technology will provide a cutting-edge AI tool to make public safety agencies inherently safer. Sensors and systems, along with connected alarms and alerts, will allow agents to perform all the necessary tasks required of them while keeping their hands on the wheel and their eyes on the road.
NEXT: How To Harness The Power Of AI In Law Enforcement