Apidays Virtual Helsinki - Merja Kajava, Aavista
summary
Pilot called Safe Cruising stitches 5G edge computing, computer vision APIs and AutoML so cars send camera frames over 5G to a nearby edge server that returns traffic‑light detections and geolocated states for other vehicles and stakeholders. Target is <5 ms latency and overall reaction budget under 150 ms (example breakdown: 20 ms capture, 20 ms upload, 100 ms CV, 10 ms return). Prototype works in trials but struggles with glare and corner cases, so custom training and local deployment are required.
AI-written summary · about
chapters (8)
- 0:00Start
- 2:265G AND EDGE COMPUTING
- 7:46CASE: SAFE CRUISING BY AAVISTA
- 8:30SCENARIO OVERVIEW
- 10:33TRAFFIC LIGHT DETECTION
- 11:33REACTION TIMES FOR ASSISTED DRIVING
- 12:03LATENCY REQUIREMENTS FOR ASSISTED DRIVING
- 14:21REQUIREMENTS FOR API
related talks
09mLightning Talk: An Efficient Automotive Safety Support System Utilizing Hardware...
44mOff to the Races with Android
22mExtending European Strength with Embedded AI
20mScaling Edge AI with Elixir and Nerves
28mVision, Big Data and AI
37mDevelop modern connected applications on 5G MEC/Edge Space and Azure
48mThe Power of Edge AI: Revolutionizing Future of Real-Time, Autonomous Processing
01h 56mEdge AI Systems