Adasa: A Conversational In-Vehicle Digital Assistant for Advanced Driver Assistance Features
summary
Adasa is a conversational in-vehicle assistant that links natural-language queries to live vehicle signals to explain and control lane keeping and adaptive cruise features. The prototype runs on a 2016 Lincoln Continental, trains intent classifiers with about 4,500 AMT utterances, supports 70+ intent classes, and achieves 92.5% intent accuracy, 77.6% perceived response correctness, and 1.5 s end-to-end latency. A 15-participant, real-road study over an 11-mile route shows usable real-time responses but mixed user nervousness and scenario-dependent response length expectations.
AI-written summary · about
chapters (12)
- 0:00Start
- 2:12Lane Keeping System
- 2:23Adaptive Cruise Control
- 4:22Goal for this Project
- 7:30Real-Time Processing
- 8:05Overview of Edessa
- 10:19Evaluation
- 11:26System Analysis
- 11:48Response Correctness
- 12:17End-to-End Latency
- 16:21Summarize Our Work
- 21:33Impact on Driving Performance
related talks
45mTalking to your driverless car
51mAutomated Interactive Domain-Specific Conversational Agents that Understand Huma...
38mAI-Powered Conversational Interfaces
21mUsing Advisory 3D Sound Cues to Improve Drivers' Performance and Situation Awareness
32mAI for the Masses Building AI Products that Users Trust
01h 06mAI as a Force for Innovation
17mThe Voice-First AI Overlay: Designing Conversational Co-Pilots
42mAgentic AI and the Future of AI