Our EchoDrive cognitive AV radar pushes a number of boundaries. The inspiration is primarily biological - human drivers regularly interrogate ambiguity in the driving scene by focusing sensor resources, typically eyes but also ears. When a driver is confused, greater focus and intensive listening provide highly-current/highly-local information for better and safer driving actions.
Today's sensor for autonomous ground vehicles offer primarily one-way data flows, from sensor through fusion to decision and actuation. But when the data is inconclusive, how can the AV stack disambiguate the driving scene?
We assert that intelligent machines must be able to to dynamically interrogate the scene, actively measuring specific areas of the driving scene in specific ways that raise probabilistic confidence in drive decision and actuation. EchoDrive is an advanced field-able prototype that offers extraordinarily rich data for machine learning and artificial intelligence development.
Designed for advanced vehicle AI and research
EchoDrive is an advanced imaging radar designed for L4+ autonomous vehicle safety.
Rich data for ML
Sensors that discard data to provide detection-level or higher data are making decisions about what data is best for training machine learning and drive decisions. That power should be in the hands of the AV stack developers. EchoDrive produces raw (pre-detection) data output that is ideal for ML-based classification and identification.
Increase probabilistic confidence
Running, walking, skateboarding, and scootering each have distinct doppler signatures, as do relative motions and cross sections of different wheel types. These data and others increase positive outcomes in drive decision and, ultimately, roadway safety.
Proven, reliable, ready to experiment
Echodyne knows how to make radars that perform. EchoDrive is an experimental, taskable, imaging, hybrid ESA/MIMO radar that pushes the boundaries of contemporary vehicle intelligence.