← Back to Publications

Neurosymbolic hybrid approach to driver collision warning

Authors:

Year: 2022
Venue: arXiv:2203.15076
Links: arXiv abstract

Summary

Combines deep-learning object recognition and tracking with an adaptive neurosymbolic NARS agent for driver collision warning. The paper uses NARS as a reasoning layer to build concepts from perceptual sequences and issue warnings in a safety-critical perception/action context.

Relevance to Hyperon

This is a practical NARS application source for MeTTa-NARS Deep Dive and a useful boundary example for neurosymbolic perception. It should be treated as NARS/ONA lineage, not as direct evidence for current MeTTa-NARS implementation status.

Key References

• RawData source
• MeTTa-NARS Deep Dive