Increase horizon of perception and facilitate decision-making collaboration in V2X scenarios

The paradigm of commuting vehicles is nowadays changing and evolving. We see a constant increase on driving assistance systems and huge efforts are being made for deploying Autonomous Driving. 

A vehicle’s ability to perceive the world around itself to be able to take decisions on how to respond and react to it is paramount for achieving this technology. Each vehicle, however, has a perception limited by the field of view of their sensors. 

One of the key points to achieve good results in this field will be the ability to perceive information beyond the field of view of the own sensors, and that is where information sharing comes in, allowing to achieve an integrated perception of the environment, as complete and comprehensive as the entities reachable through the network. 

 

OUR APPROACH 

Our innovation is to make use of Roadside Units (RSUs) as lightweight edge computing devices for sense fusion and lightweight storage and computation. The aim is to provide a real-time (distributed) service without costly Cloud or heavy Edge solutions. 

 This is how we do it: 

  • Run distributed (replicated) Services, e.g., traffic density, obstacle detection, on the RSU edge network. 
  • Provide real-time updates on corresponding client applications on vehicles’ On-Board Units (OBUs)
  • The RSU network is loosely connected; we allow for real time concurrent reads and writes on RSUs to avoid synchronization overhead.
  • Use Conflict-free Replicated Data Types (CRDTs) to ensure data convergence despite the potential conflicts due to concurrent updates 

 

 

KEY BENEFITS  

  • Low latency: avoid Vehicle-Cloud and Vehicle-Edge roundtrip communication delays 
  • Low cost: no Cloud/Edge subscriptions required 
  • Low network overhead: avoid pushing transient data to the cloud 
  • Privacy: no need to push local/city data to Cloud
  • Integrate as a separate edge layer complementary to Edge-Cloud ecosystem 
  • Useful for plenty of AD use-case scenarios: Live HD Maps, Robust Platooning, Collaborative detection, accurate Geo-location, Second Brain 
  • Seamless integration of shared information in the vehicle’s own environment representation 

Documents