The AE110 Brings Together Agile Sensors, Artificial Intelligence, and Software Definability to Capture Better Information with Less Power, Raising the Bar on Performance and Safety
“This is a breakthrough for perception systems. iDAR can do what has been previously unachievable. Combining industry leading range, resolution, scan rate, and intelligence – we are able to capture critical data about objects such as acceleration and velocity vectors within the same frame. Our approach has superior characteristics to simple FMCW not only in reliability of acquiring closing velocity, but also as the most dangerous objects enter the scene from the side where FMCW alone is ineffective.”
Pleasanton, CA – December 20, 2018 – AEye, a world leader in artificial perception systems and the developer of iDAR™, today announced the AE110, its most advanced product for the automotive mobility market. Built on the world’s most extensive solid-state LiDAR patent portfolio (with over 1000 claims filed), the AE110 features the industry’s only software-definable LiDAR, creating an open platform for perception innovation that lets software engineers optimize data collection to best meet their needs.
A New Kind of Perception
The AE110 is based on iDAR, a new form of data collection that emulates how the visual cortex pre-processes and customizes information sent to the brain. It does this by fusing 1550 nanometer (nm), solid-state agile MOEMS LiDAR, a low-light HD camera, and embedded AI to intelligently capture data at the sensor level. iDAR integrates 2D camera pixels and 3D LiDAR voxels at the point of data acquisition to create “true color” Dynamic Vixels – a new real-time sensor data type that delivers more accurate, longer range and more intelligent information faster to the car’s “brain”, or its path-planning system.
The product has two major architectural advantages: the camera and LiDAR are fused and boresighted, and it’s the only LiDAR system that’s entirely software-defined. As a result, the AE110 delivers high quality information faster to path planning systems, using significantly less power. Further, its straightforward API and powerful SDK transforms iDAR into a unique open system for perception innovation, allowing perception engineers to easily customize and dynamically control how scenes are interrogated. In addition, the product is under ASIL-B functional safety certification, ensuring it meets the automotive industry’s stringent safety standards.
“The AE110 sets a new bar for what perception systems are capable of,” said Aravind Ratnam, VP of product management at AEye. “By using an agile, software-definable LiDAR, the AE110 is able to collect 4-8 times the information of a fixed pattern LiDAR, using an eighth of the points. If a child walks into a scene, we can revisit that image within microseconds to help the vehicle take evasive action. Efficiency, speed and accuracy are key to perception, and the AE110 delivers all three at unprecedented levels at an extremely attractive price point.”
Key features of the AE110 include:
- Software-definable LiDAR with high resolution RGB (red, green & blue) camera mechanically fused to deliver both true color and geometry as Dynamic Vixels
- High resolution RGB camera with high speed interface
- Onboard computer vision and embedded feedback loops for enhanced machine learning
- Designed to meet ISO 26262 functional safety requirements; eye safe in all modes
- Automotive SWaP optimized with documented field and environmental testing
- Software Development Kit
- Automotive deployment optimized with minimal crosstalk, utilizing four layers of interference mitigation and anti-spoofing technology
- Enhanced adverse weather capabilities
The AE110’s unique architecture provides for the highest performance in resolution, range and refresh rate per dollar, while delivering unparalleled power savings. For example, its pseudo-random beam distribution search option makes the system eight times more efficient than fixed pattern LiDARs. The AE110 also achieves 16 times greater coverage of the entire field-of-view at 10 times the frame rate (up to 100 Hz) due to its ability to support multiple regions of interest for both LiDAR and camera.
Conventional LiDAR systems are designed to provide basic search capabilities – passively capturing limited data and treating it all equally without distinguishing critical threats or objects in the scene. In contrast, AEye’s iDAR performs multi-modal intelligent search which can then acquire, pre-classify, and track multiple objects quickly. It’s also designed to allow the system to pre-classify the objects faster than the frame-rate by allowing for a much more rapid object revisit rate.
“This is a breakthrough for perception systems,” said Allan Steinhardt, Chief Scientist at AEye and former Chief Scientist at DARPA. “iDAR can do what has been previously unachievable. Combining industry leading range, resolution, scan rate, and intelligence – we are able to capture critical data about objects such as acceleration and velocity vectors within the same frame. Our approach has superior characteristics to simple FMCW not only in reliability of acquiring closing velocity, but also as the most dangerous objects enter the scene from the side where FMCW alone is ineffective.”
Setting New Benchmarks
Recently, AEye announced that its iDAR system set a new benchmark for both range and scan rate performance for automotive-grade LiDAR. In a test monitored and validated by VSI Labs, The system detected and tracked a truck at 1,000 meters (or one kilometer). AEye also achieved a scan rate of 100Hz, setting a new speed record for the industry. This comes on the heels of the close of AEye’s series B funding, led by Taiwania Capital, bringing the company’s total funding to over $61 million.
The AE110 will be available to automotive and mobility OEMs and Tier 1s in Spring, 2019. For more information about AEye, the AE110, and the company’s innovative approach to artificial perception, please visit CES Booth #2100 at the Westgate Convention Center in Las Vegas, January 8th – 11th, 2019, For a private car demo, please contact [email protected].