Tag: Autonomous

On June 26th, AEye Staff Engineer, Vivek Thotla, will be speaking on a panel called “Should We Take CV To The Edge?” at IoT Forum on Computer Vision @ Sensors Expo.
Vivek is a staff engineer at AEye, where he leads product verification and validation, and is responsible for LiDAR simulation and data strategy in producing automotive grade products. Previously, he was a Component Owner / Functional Delivery Owner for point cloud algorithms at Continental, where he was responsible for planning, requirements, design and development of embedded platform-based algorithms for a Hi-Res 3D Flash LiDAR, in addition to enforcing ADAS process stages to meet ASPICE levels and functional safety. He has also held engineering roles at Tribis, AmpliSine Labs, Missouri S&T and Enigma Portal. Vivek holds an MBA in Information Technology Project Management and a PhD and Masters in Electrical, Electronics and Communications Engineering.

We sat down with Vivek to learn more about the advantages of integrating computer vision at the sensor, building automotive grade LiDAR products, and why he decided to move to the Bay Area.

Q: How much of an autonomous vehicle’s computer vision should be done at the sensor, as opposed to a central processor?
The amount of data produced today by a perception system is enormous. And incorporating all the data from the different kinds of sensors used (like radar, camera, and LiDAR) makes it very difficult and expensive to process and store. In a typical perception system, roughly 80% of the data produced by the sensors is thrown out.

However, intelligent sensors – like what we develop at AEye – are software definable. Meaning, you can adjust its settings to get high resolution data from an object and get sparse data in the background, cutting down the overall amount of data processed by more than 80%. This makes computer vision algorithms at the central processor faster and efficient because once you preprocess data, latency becomes less of an issue. Currently, AV companies are spending a tremendous amount of money storing useless data. Preprocessing saves both time and money.

Q: What is the largest challenge in producing automotive grade LiDAR products?
Industry wide, the greatest challenge is maintaining the quality, reliability, and consistency needed on all components and software that go into a LiDAR sensor of over 100,000 samples or more and over the sensor’s lifetime. Another major challenge for bringing LiDAR products to the automotive market is designing the sensor to fit in different regions of the car. There are a lot of constraints based on where the sensor is placed on the vehicle and certain issues that arise from each placement. For example, a sensor placed behind a windshield might need a completely different design than a sensor that’s placed in the front bumper.

There are many interesting LiDAR architectures out there that work really well at smaller samples and in the lab. But the moment the product needs to scale and deal with all the quality and environmental requirements of being an automotive grade product, they fail. AEye is mitigating these challenges by partnering directly with Tier 1’s who know the process of making large-scale, automotive grade products. In my own experience, I’ve found that once a Tier 1 partners with you, they are extremely supportive because they believe in you, and that proves you are capable of achieving it.

In addition to our partners who help us push the sensor to automotive grade, we have a great functional safety team here at AEye. I came to AEye from a Tier 1, so I know what goes into developing an automotive grade sensor, and the AEye team is made up of people from all over the automotive industry that have great, diverse insight into how to bring a product to market.

Q: You moved to the Bay Area from Santa Barbara. What was it about Silicon Valley that drew you here?
It has always been my dream to come to Silicon Valley – you hear about it so much as the epicenter of technology and innovation. And it’s true: Silicon Valley is at the heart of the autonomous driving industry. All the innovative and novel work happening today in the LiDAR industry is happening here and I did not want to miss my chance to help develop the tools for true autonomy.

Connect with AEye at Sensors Expo! Learn more here.

AEye Team Profile: Vivek Thotla — VentureBeat Says iDAR Is “Built for Speed”AEye Team Profile: Dr. Allan SteinhardtAEye Adds VP of AI and Software to Executive TeamAEye Advisory Board Profile: Scott PfotenhauerThe Future of Autonomous Vehicles: Part II – Blind Technology without Compassion Is RuthlessAEye Advisory Board Profile: Adrian KaehlerAEye Team Profile: Indu VijayanAEye Expands Business Development and Customer Success Team to Support Growing Network of Global Partners and CustomersAEye Sets New Benchmark for LiDAR Range