Follow That Car began with a simple observation: people often explain a route by pointing to something in the world—“follow that car”, “turn after the church” or “take the road beside the park”—while navigation systems largely communicate through abstract instructions.
This project investigates whether nearby traffic can become part of the navigation interface itself.
Visual navigation in action
Follow That Car identifies a vehicle that is already taking the intended route and presents it as a temporary guide.
The clip includes the system's spoken advice, so switch the sound on.
A recorded replay of the Follow That Car prototype combining camera imagery, route-relative reasoning, guide selection and spoken advice.
Observe
A stereo camera detects and tracks vehicles in the traffic ahead.
Relate
Detected vehicles are positioned relative to the planned route and upcoming maneuver.
Reason
Symbolic rules determine which vehicles are plausible guides, competitors or unsuitable candidates.
Advise—or don’t
The most distinctive suitable guide is described, unless silence is the safer or clearer outcome.
Follow That Car combines learned perception with explicit symbolic reasoning. Computer vision supplies observations about the traffic scene. Route-relative geometry gives those observations spatial meaning. Answer Set Programming selects a guide—or determines that no reliable advice should be given.
The prototype is developed and evaluated as part of a Master of Science thesis in Computer Science.
Follow That Car is evaluated across recorded urban driving runs, route providers, positioning sources, maneuver types and alternative system configurations.
