Bay Vision - Spring 2018

software was optimizing the process of collecting intelligent information, and in doing so, allowing a lot of software configurability within the actual sensor itself. Probably one of the hardest elements to introduce to customers was the ability to have full control or full flexibility within a system – Customers were so deeply rooted with the hardware they were familiar with that convincing them that there was a whole new paradigm of capability was a challenge in itself. Jordan tells us there are two things he is most proud of: “ The first is that in order to get this unique data set that we were after, we had to create novel hardware and the first component of that novel hardware was what we called an agile LiDAR. The agile LiDAR is a LiDAR that’s extremely fast and can scan any pixel within the field of view. We can direct the laser to interrogate any point in space, which means that with the complementary camera data, we can enable new capability and unmatched performance in the perception stack. The second is the fact that we were the first, and still I believe, the only system that has been able to seamlessly stitch together multiple solid- state systems on top of a vehicle for 360- degree coverage . ” In January, AEye announced their AE100 product , which is the automotive product that is production-ready and scalable, intended to address mobility markets. Jordan says the future is looking very interesting, because their platform is extensible: “ It’s actually a plug and play model where there is a lot of value that we can extract from orthogonal data sets. Convolutional neural networks in particular thrive on orthogonal data sets. Our platform enables other sensors such as radars, IMUs, ultrasonics, microphones and various other vehicle instruments to optimize our data collection .” He goes on to explain that the future is really not in the hardware, but the software. If someone comes out with a better subsystem, whether it’s a LiDAR camera or radar, AEye would be happy to swap it in or figure out how to optimize the way that they collect the information for the path- planning software. The future resides within the very low latent algorithms that they have today, as well as the perception level software that they develop on top of that to optimize the way that they can improve the data sets that are being delivered to the path-planning software. The company wants to deliver the highest quality information to path-planning software to ensure that the cars or vehicles on the road are not just relying on the base data, but actually intelligent perception . 7 AEye Bay Vision