AI Online


Innoviz Technologies: Pioneering Autonomous Driving with InnovizCore AI Compute Module and Strategic Collaborations in the Automotive Industry

In the fast-evolving landscape of autonomous driving, Innoviz Technologies stands at the forefront, pushing the boundaries of innovation. Recently named a CES 2024 Innovation Awards Honoree for their groundbreaking InnovizCore AI Compute Module, we had the privilege of sitting down with David Oberman, VP Asia at Innoviz Technologies, to dive deep into the remarkable developments that are shaping the automotive industry.


InnovizCore AI Compute Module: Redefining Autonomy


Congratulations on being named a CES 2024 Innovation Awards Honoree for the InnovizCore AI Compute Module. Oberman begins by shedding light on the significance of this module in advancing autonomous vehicle technology. As a Tier1 automotive supplier, Innoviz now offers a comprehensive solution, comprising the Optical Module (LiDAR) and the Compute Module (InnovizCore or “iCore”). This end-to-end approach allows for tailored solutions, aligning with the unique needs of major automotive OEMs.


The iCore, a robust and flexible Compute Module, integrates inputs from multiple sensors, including LiDARs and Cameras, producing an integrated point cloud image and performing advanced Perception calculations. Running on a Qualcomm Snapdragon Ride SoC with AI and GPU hardware, the iCore is equipped with massive computational accelerators capable of unprecedented results. What sets it apart is its role as a backup Autonomous Driving processor. In the event of a fault in the central ECU, the Innoviz iCore seamlessly takes control, ensuring a smooth and safe stop – a groundbreaking capability in the industry.


LiDAR’s Role in Autonomous Driving: Overcoming Challenges


Moving on to LiDAR’s role in enabling autonomous driving, Oberman emphasizes the crucial importance of this technology, particularly from Level 3 autonomy onwards. LiDARs, when used in conjunction with cameras and radars, provide a comprehensive solution for autonomous driving. Innoviz, with its market-leading resolution and dense point cloud, tackles challenges such as 3D vision, night-time visibility, and adverse weather conditions. The LiDARs are designed to be robust, particularly excelling in rain, where visibility remains unaffected.


Strategic Partnerships with BMW and Volkswagen: Catalysts for Success


Innoviz’s strategic partnerships with automotive giants like BMW and Volkswagen have proven instrumental. Oberman acknowledges these collaborations as transformative, with BMW and the VW Group serving as invaluable partners and mentors. The nominations for mass production programs and huge orders have propelled Innoviz into a position of strength, thriving in a competitive market where others face struggles.


Scalability and Cost-Effectiveness: Key Drivers of Success


Known for producing compact, scalable, and cost-effective LiDAR solutions, Innoviz focuses on constant improvement in performance while concurrently reducing costs. Oberman emphasizes the importance of efficiency in design for future generations of products, recognizing the constraints of limited space, power consumption, and competitive pricing in the automotive industry. The “can-do” attitude of Innoviz’s technical team, numbering nearly 500, is highlighted as a driving force behind their success.


AI-Powered Perception Software: The “Eyes” and “Brain” of Autonomous Vehicles


The synergy between Innoviz’s LiDAR sensors and AI-powered Perception Software is aptly described as the “eyes” and “brain” of autonomous vehicles. LiDAR emerges as the key sensor for driving decisions, with its 3D sensing ability, precise distance measurements, and day-and-night effectiveness. The Perception Software, or Computer Vision, plays a critical role in processing the point cloud image, providing essential information to the central ECU for real-time decision-making.


Advancements Towards Levels 3 and 4 Autonomy: Realizing the Dream


Innoviz is actively advancing towards making Levels 3 and 4 autonomy a reality. Oberman expresses excitement about the fruition of their Level 3 program with BMW, marking a significant milestone. The company is also supplying LiDARs for Level 4 shuttle programs and aiming for nominations in Robotaxi fleets. While Level 5 autonomy may take time to achieve, the progress made by Innoviz and other industry players signals a positive trajectory for the future of autonomous driving.


The full interview with David Oberman provides a captivating insight into Innoviz Technologies’ pivotal role in shaping the automotive industry’s autonomous future. Their commitment to innovation, strategic collaborations, and a relentless pursuit of scalability and cost-effectiveness position them as leaders in a dynamic and evolving landscape. As the autonomous driving industry emerges from its “trough of disillusionment,” Innoviz Technologies stands poised to drive real-world progress and contribute to the widespread adoption of autonomous vehicles globally.


Automotive industries InnovizCore AI Compute Module Recognition: Congratulations on being named a CES 2024 Innovation Awards Honoree for the InnovizCore AI Compute Module. Could you elaborate on how this module contributes to advancing autonomous vehicle technology, and what sets it apart from other solutions in the market?


Oberman: This new development is closely connected with Innoviz’s becoming a Tier1 in 2022, in the automotive industry.   Innoviz develops both the LiDAR itself (what we call the “Optical Module”) as well as the Perception Software, also known as “Computer Vision”, which often runs on a separate ECU, which is called the “Compute Module”.  When Innoviz was a Tier2 company, we provided our Perception Software to our Tier1 partners, who built the Compute Module.  Now that we are an automotive Tier1, we are offering a complete suite of solutions, providing an end-to-end solution with both the Optical and Compute Modules, tailoring our solutions to the needs of each big automotive OEM.


Our Compute Module, the InnovizCore or “iCore”, is built to be a very strong, robust and flexible solution, accommodating inputs from multiple sensors such as LiDARs and Cameras to produce an integrated point cloud image, and performing Perception calculations.  The iCore also gets CAN signal inputs from the car.


The iCore Compute Module from Innoviz runs on a Qualcomm Snapdragon Ride SoC with advanced AI and GPU HW, and uses a Marvel network switch.  Integration of this ethernet switch enables the InnovizCore to manage the vehicle network – reducing the complexity of vehicle network management, and saving hardware cost.  The iCore includes massive computational hardware accelerators, (CNN, GPU, DSP), which are capable of achieving unprecedented results achieved through its Perception capabilities.


One of the unique points about Innoviz’s iCore Compute Module is that it acts as a back-up Autonomous Driving processor for the car, and in case of any fault in the central ECU of the car, the Innoviz iCore takes over the driving of the car, and performs MRM, a Minimal Risk Manoeuver, or in other words, performing a smooth and safe stop.  This is, as far as we know, the first time that a processor provided by a LiDAR manufacturer is capable of taking full control of the vehicle unassisted.


Automotive industries LiDAR’s Role in Enabling Autonomous Driving: Innoviz is at the forefront of enabling mass production of autonomous vehicles through automotive-grade LiDAR sensors. How do these LiDAR sensors contribute to overcoming challenges in autonomous driving, particularly in terms of reliability and performance in diverse weather conditions?


Oberman: It has long been accepted that LiDAR is needed for Autonomous Driving, at least from Level3,  i.e. autonomous driving on the highway.  However, LiDAR is increasingly being adopted by OEMs also for Level2+ Advanced Driver Assistance Systems (ADAS).  LiDARs are used together with cameras and radars for autonomous driving solutions.  The main advantages of the LiDAR over cameras are its 3D vision capability and its ability to “see” just as well in dark/night-time/lowlight road conditions, as in daytime driving.  The main advantage of the LiDAR over radar technology is its high angular resolution, and on this point, Innoviz has a huge advantage, due to its market-leading resolution and extremely dense point cloud.


Making an Automotive Grade LiDAR means years of hard work developing, verifying, validating, testing and certifying the LiDAR.  The LiDAR must have the highest quality levels, and must be robust enough for 15 years operation in the vehicle.  As Innoviz has been working with Tier1 and OEM partners since January 2016, our standards have been driven by the needs of our partners and customers, and we have accumulated a wealth of knowledge in multiple disciplines.  Innoviz probably has more experience than any other LiDAR company in this respect.


Regarding diverse weather conditions:  Our LiDARs have been developed to be particularly hardy in rain, and the rain has only a negligible effect on visibility.  Our OEM customers are very surprised when they first see this characteristic.  Particularly heavy fog is challenging for all cameras and LiDARs, and the entire industry is still looking for a solution to heavy fog.  Currently the solution is the same as that for a human driver:  If you are in a “pea-souper”, don’t drive.  Stop by the side of the road until you can see the road ahead.  Cessation of Autonomous Driving under heavy fog, and driver take-over within x seconds, is already programmed into Autonomous Cars today.


Automotive industries Collaboration with BMW and Volkswagen: Innoviz has strategic partnerships with automotive giants such as BMW and Volkswagen. How do these collaborations influence the development and implementation of Innoviz’s LiDAR technology in real-world automotive scenarios? Are there any specific challenges or breakthroughs that have emerged from these partnerships?


Oberman: Innoviz’s close working relationships with BMW and the VW Group have given the company huge advantages.  These OEMs have been the best partners and teachers for us, and have helped us achieve our status as Tier1 automotive supplier.  In addition, having nominations for multiple mass production programs, with huge orders, has helped us to achieve economies of scale.  The BMW, VW Group and other OEM programs have helped Innoviz thrive in a very competitive market where other LiDAR providers are struggling to survive, and mergers and acquisitions as well as bankruptcies are the order of the day.


Automotive industries Scalability and Cost-Effectiveness: Innoviz is known for producing compact, scalable, and cost-effective LiDAR solutions. Can you share insights into the scalability of your LiDAR technology and how Innoviz addresses the challenge of making autonomous driving technology more accessible and affordable for mass production?


Oberman: Through many generations of product, starting with a prototype in 2016, InnovizPro in 2018, InnovizOne in 2020 and InnovizTwo in 2022, Innoviz has constantly improved performance, while managing to reduce cost in parallel.  This combination of improved performance and reduced cost and price, has proven to be the golden key to winning OEM mass production programs.  We are constantly improving performance and pushing down costs through efficient design of our next generations of products.  We understand well that the “real estate” space in a car is limited, power consumption is limited (especially in electric vehicles), and the price of any part of a vehicle is super-competitive, as the price must be borne by the end-user.  This means that product design must give best performance, the product must use as little power as possible, the sensor must be as small as possible, and the price must be competitive and meet the OEMs’ price targets.


Innoviz has a very open approach to the technical needs of each automotive manufacturer, and makes extra efforts to accommodate the different needs of each company.


The key to achieving all of the above is the “can-do” attitude of our nearly 500-person strong technical team, who constantly look for ways to improve every aspect of our products.  Our engineers’ motto is: “No challenge, no fun!”


Automotive industries AI-Powered Perception Software: The combination of Innoviz’s LiDAR sensors and AI-powered Perception Software is referred to as the “eyes” and “brain” of autonomous vehicles. How does this synergy enhance the vehicle’s ability to perceive and navigate its surroundings, and what role does the AI-powered software play in making real-time decisions for safe autonomous driving?


Oberman: The LiDAR is the key sensor for making driving decisions.  It is the most capable sensor due to its 3D sensing ability, its ability to know the exact distance to each and every object in the road, and its ability to work equally well by day or by night.  A camera is a 2D, passive sensor.  It does not know the exact distance to any car, truck, person, motorbike, bicycle or obstacle on the road.  A LiDAR knows the EXACT distance to any such object, measuring it 10 or 20 times a second. The Perception Software or Computer Vision, which detects, in the point cloud image created by the LiDAR, everything on the road, and distinguishes between various classes of objects, is the part of the system which gives the information needed to the central ECU.  Then the ECU can make sure that the vehicle can stay in its lane, slow down to keep distance from the vehicle ahead, stop, change lanes, and execute emergency braking when needed.  This kind of AI is certainly one of the most important applications for mankind today, and is already responsible for saving lives on the roads of the world.  In a few years, we hope that it will be responsible for reducing the number of people killed and injured on the roads of the world, by around 90%.


Automotive industries Advancements Towards Levels 3 and 4 Autonomy:  Innoviz is actively working towards making Levels 3 and 4 autonomy a reality. Can you provide insights into the specific technological advancements or breakthroughs that Innoviz is focusing on to achieve higher levels of autonomy? Additionally, how do you see the landscape of autonomous driving evolving in the next few years?


Oberman: We are thrilled to see our first Level3 program with BMW, for the BMWi7, come to fruition this year.  The BMW cars are on sale now and will be on the roads in Germany from March this year, and later in China in the form of the BMW5.  For us, this is our dream come true, supplying the most sophisticated and cutting-edge sensor technology with AI-based Software, for industry-leading BMW’s top cars.  We are also supplying our LiDARs for Level 4 shuttle programs, and trying to win nominations for Robotaxi fleets.   Level5, or complete autonomy, will take a good few years to achieve.


In general, I think that we are now emerging from the Autonomous Driving technology’s “trough of disillusionment” and starting to make real progress on autonomous driving programs. Pacesetters like BMW and VW have ensured that Autonomous Driving is here to stay, and the number of new car models with LiDAR is growing and will continue to grow each year .  LiDARs can already be seen on many new car models in China, from carmakers such as NIO, Xpeng and Li Auto, among others, and it will not be long before autonomous cars with LiDARs are ubiquitous in North America, Europe and large parts of Asia.


Previous posts

Next posts

Sun. July 14th, 2024

Share this post