Practising the moment to step in.
Immersive XR training for the ABLE Project (Active Bystandership for Law Enforcement) at Georgetown Law, which prepares police officers to step in when a colleague is about to cause harm. Made with Meta CS Labs, 1stAveMachine and SpecialGuestX; we led the AI and XR development.
Three cases put the trainee inside the scene: Sgt. McDonald, after an arrest at a domestic shooting; Officer Lopez, at night in a small station; and a traffic stop that escalates. The characters speak to the trainee and wait for an answer, and the case changes with it. The scenes combine real actors, AI-based motion tracking and facial animation, volumetric capture and environments modelled in Unreal Engine. The first builds were tested with officers in Washington DC in July 2023, and the experience was shown at the Knoxville Police Department that December.
Capture
Real actors are recorded and tracked with AI, and their faces are animated from real audio.
Recreate
The characters become MetaHumans or volumetric captures, placed in scenes modelled in Unreal Engine.
Light
Lighting and camera are tuned case by case, scene by scene.
Step in
Officers go through each case in VR, answer out loud and practise the moment to intervene.
Text to XR
Within ABLExR we ran a text-to-XR research track: a written prompt becomes a character animation, is retargeted to a MetaHuman and placed in a scene ready for VR or AR.
Working on something like this? Tell us what you need.
