AEye's iDAR Robotic Perception for Autonomous Vehicles
Company: AEye, Pleasanton, CA
Company Description: AEye is a robotic perception pioneer whose advanced vision hardware, software & algorithms act as the eyes and visual cortex of autonomous vehicles. The company invented iDAR, a new form of intelligent data collection that enables autonomous vehicles to not just capture a scene, but to dynamically perceive it, just like a human would - honing in on critical information that saves lives.
Nomination Category: Company / Organization Categories
Nomination Sub Category: Technical Innovation of the Year - At Organizations With Up to 100 Employees
Nomination Title: AEye's iDAR Robotic Perception for Autonomous Vehicles
Tell the story about this nominated innovation since January 1 2017 (up to 650 words). Describe how the innovation was discovered, developed, refined and/or deployed. Provide an assessment of how the innovation has impacted or will impact your organization, your industry, markets or society.
iDAR™ came about to solve the challenge of cars being able to drive themselves. AEye’s founder and CEO hails from the defense industry, where he worked at places like NASA, Northrop Grumman, and Lockheed Martin on both really large aircraft surveillance and targeting systems, as well as lightweight compact reconnaissance systems for infantry. His experience convinced him that the perception stack was the key to machine vision, and that to achieve the performance of AI to do tasks, the automotive industry would need a perception engine that could meet or exceed a human’s ability to perceive the environment.
While many companies were looking to LiDAR (Light Detection and Ranging) as the elixir to enable robotic vision, AEye knew that LiDAR alone was not enough. Because it lacks intelligence, a LiDAR-only solution is unable to take into account how a scene evolves or what the mission is. It simply collectsas much data as possible, without discretion, and passes it up to a CPU, where 75-95% of the collected data is discarded because it's redundant or useless. This creates a huge strain on bandwidth, causing latency.
AEye created iDAR (Intelligent Detection and Ranging) to bring intelligence to the data collection process. iDAR fuses a low cost solid-state agile LiDAR with best of breed camera tech and embedded AI to create software-definable hardware that can dynamically adapt to real-time demands. In other words, iDAR enables a machine to dynamically interrogate a scene to focus on what's relevant.
By enfusing agile targeting and intelligence into the data collection process, iDAR is able to immediately start throwing away or triaging data in order to perform decision analytics. This allows the system to capture more information with less data, enabling much faster perception and path planning - a key to the safe rollout of autonomous vehicles.
In December 2017, AEye introduced iDAR, as well as its iDAR Development Partner Program for OEM customers, Tier 1 partners, and universities interested in integrating iDAR into their vehicles. Its first iDAR-based product, the AE100 Robotic Perception System, was announced in January 2018, and will be available this summer to OEMs and Tier 1s launching autonomous vehicle initiatives. In February 2018, the automotive industry voted AEye’s iDAR as the Best Newly Offered LiDAR Solution.
This innovation will have an enormous impact on the autonomous vehicle sector. With IDAR, for the first time a machine is able to mimic how a human’s visual cortex focuses on and evaluates potential driving hazards. Instead of giving equal weight to everything in a scene, iDAR uses agility (no fixed scan pattern), a distributed architecture, and at-the-edge processing to dynamically track targets and objects of interest (i.e. a child) while always critically assessing general surroundings (i.e. the sky road).
Adding agility and intelligence to the sensor layer means objects of interest can be identified and tracked with minimal computational latency - that translates into faster and better predictability of behavior and intent. Also, by adding computer vision to 3D point clouds, perception and path planning software can now leverage thousands of existing image processing algorithms for much greater accuracy and speed in interpreting signage, emergency warning lights, brake vs. reverse or traffic lights, and other scenarios that have historically been tricky for LiDAR-only systems to navigate. And because its software-definable, an iDAR system can adapt to environments as they change, such as city versus highway driving.
With iDAR, for the first time in history, the mass market will have access to powerful sensing capabilities rivaling aerospace and defense capabilities. This is a major technological breakthrough, and one that promises to change the way all robots, big and small, perceive the environment.
In bullet-list form, briefly summarize up to ten (10) chief characteristics or benefits of the nominated innovation.
-Enables machines (cars) to go beyond seeing to actually perceive their environment
-Infuses real-time AI into the data collection process to collect relevant vs. all data
-Eliminates the collection of redundant and useless information
-Frees up bandwidth to perform more sophisticated AI computations
-Delivers higher accuracy, longer range, and more intelligent information to optimize path planning software
-Enables rapid, dynamic perception and path planning - which speeds a car’s time to reaction
-Open, extensible platform helps mobility players streamline their development process and bring better autonomous vehicles to market, faster