A number of companies are trying to crack the autonomous driving puzzle, and a variety of approaches have evolved. Some companies are taking a software-first approach, building an AD software stack that can work on any hardware environment. Others are taking a hardware-first approach, creating a sensor and hardware environment that can adopt any software stack. Others still are creating a ‘walled-garden’ with software and hardware designed in close conjunction, allowing each to work only with the other for a specific use case. Imperium Drive, a UK-based autonomous driving startup, has a radically different take on the autonomous driving problem. Imperium Drive believes that a ‘human-in-the-loop’ is a critical stop on the journey to full autonomous driving. The company is focused on bringing commercially and operationally viable products to the market, even as it pursues its ultimate goal of making autonomy a reality.We caught up with Koosha Kaveh, CEO of Imperium Drive on this episode of the AI in Automotive Podcast. Koosha shares his view of the evolution of AD, and progress on the autonomy journey. He also introduces us to Fetchcar, their super interesting driverless, human-in-the-loop car rental service.I hope you enjoy listening to this episode, and if you do, please do share it with your network on LinkedIn and rate our show wherever you get your podcasts.https://www.ai-in-automotive.com/aiia/120/kooshakavehAI in Automotive Podcast

The digital screens in the car are the latest battleground in the battle for the driver’s attention. The main competitor? The mobile phone. Each time you choose to connect your smartphone to your car’s infotainment system using CarPlay or Android Auto, the mobile phone wins. The car’s digital surfaces are reduced to dumb screens that mirror the phone’s apps and use the phone’s processing power. Your usage data and analytics are no longer accessible to the car maker, putting them at a significant disadvantage.So, can car makers, famously still on the software learning curve, compete with the Apples and Googles of the world - companies that build software for a living - in this software-led automotive world?In this episode of the AI in Automotive Podcast, we are hosting Alex Barth, VP of Automotive and Mobility at Mapbox. Mapbox is a mapping and location cloud platform that offers the building blocks; the SDKs and APIs to power navigation. Their new product, Mapbox Dash, helps car manufacturers create brand-specific in-car digital experiences anchored to navigation, and generate subscription revenue. Alex shares some amazing insights on how automotive manufacturers can use the power of AI to deliver compelling in-vehicle digital experiences. AI in Automotive Podcast

90% of all the data in the world was generated in the last 2 years. I have read this statistic in a news report from 2012, and a news report from 2019 and a news report from 2021. If this statement is true every year, you can begin to imagine the rate at which data generation and use is expanding. Actually that number is so big, you probably can’t imagine it. As someone famous once remarked: with great data comes great responsibility. Infinite amounts of data raise questions about the ownership of data, the uses of data and the ethics surrounding data. These are all themes we explore with Richard Barlow, Founder & CEO of wejo. Wejo is on a mission to be the data ecosystem and the comms stack for the automotive industry. Richard brings some really powerful counter-intuitive perspectives to our discussion, that will challenge all the cliches you have heard about data, data usage, data ownership and data security.If you liked my chat with Richard on this episode of the AI in Automotive Podcast, please do share this episode with your colleagues and connections, and be sure to rate us on Apple Podcasts.AI in Automotive Podcast

Automotive product development is not just incredibly expensive, but also an extraordinarily complex process. Everything you see on a production vehicle is, at the end of the day, the result of hundreds of tug-of-war contests. Design and cost, manufacturability and serviceability, power and weight. The list goes on. Dozens of teams with hundreds of engineers pursuing conflicting design objectives, but eventually finding an optimum through a series of very difficult trade-offs.Any benefit an automaker can take in the design and development process that accelerates take-to-market and improves margins is an absolute no-brainer. So what role can machine learning play?In this episode of the AI in Automotive Podcast, I am joined by Dr Richard Ahlfeld, CEO of Monolith. We discuss how machine learning can be a potential game-changer in automotive R&D, and deliver some serious acceleration and cost benefits in the design and development process. Complex mathematical equations, ketchup bottles and a million dollar prize. This episode of the show has it all!AI in Automotive Podcast

In this episode of the AI in Automotive Podcast, we are following the incredible journey of Ocado from a retailer to a global technology business that offers a cutting-edge technology platform to other retailers. We talk to David Sharp, Head of Head of Autonomous Mobility at Ocado Technology, and he talks to us about their journey - bold and audacious in more ways than one.Three things really stood out for me from this very enlightening conversation with David.One, the way Ocado experiments with, integrates into and scales AI technology for various use cases in their business.Two, the way Ocado is using the composite of traditional statistics, new generation AI and time-tested human intelligence to deliver dramatic order-of-magnitude impact in their business. Three, the autonomy experiments Ocado is running in moving goods from their warehouses to their customers, and what lessons might that have for the larger automotive and mobility industries.Ocado’s journey to this point, and their ambition for the future offers so much inspiration for the larger automotive industry. Will we ever see an automotive retailer create a technology stack that they sell to OEMs and other retailers? Never say never!AI in Automotive Podcast

Have you ever thought about all the things that go on behind the scenes when you file an insurance claim for your car? Probably not. And that’s a good thing, because if you did have to think about the complexity that lies behind, you would not be a happy customer.The reality, however, is that every insurance claim triggers hundreds of processes and transactions that work in harmony to verify and pay out your claim. The fascinating thing is that AI is starting to play an increasingly influential role in this ecosystem.To understand this better, we invited Marc Fredman to the AI in Automotive Podcast. Marc is the Chief Strategy Officer at CCC Intelligent Solutions. CCC’s technology connects thousands of companies that make the insurance economy run and powers over a hundred billion dollars of insurance commerce.In today’s conversation, Marc takes us inside the insurance economy, and shares with us the variety of ways in which artificial intelligence is helping power the insurance ecosystem.If you like this episode of our podcast, do share it with your friends and colleagues, and rate us wherever you get your podcasts.AI in Automotive Podcast

If Elon Musk says something is hard, you can bet your bottom dollar that it is bloody hard. Automotive applications are some of the toughest, most unforgiving environments to deploy AI in. Most AI applications in automotive are mission critical - they just can’t go wrong. This requires that all the input they process is of a very high resolution or quality. They also need to operate in real-time, often perceive the environment around them and respond to it in a matter of milliseconds. Lastly, they need to operate with a very low power requirement, in an environment that is subject to dust, vibration and wide temperature variations. This intersection of constraints means that deep learning algorithms for automotive applications need to be designed right, and optimised constantly. This is where Deci comes in. Think of Deci as an easy-to-use platform that offers a set of tools to optimise your machine learning algorithms, and deliver order-of-magnitude performance improvements. In this episode of the AI in Automotive Podcast, I am joined by Yonatan Giefman, Co-founder & CEO of Deci. Yonatan tells us how AI applications are brought to life, and why they need to be optimised. We discuss the Deci platform, and how it can help data scientists and deep learning architects to optimise their algorithms, using the company’s proprietary neural algorithm search technology. We also talk about machines building machines, and even take a peek into science fiction :)If you liked my chat with Yonatan today, do subscribe to the AI in Automotive podcast, and give us a shout on your social media.https://www.ai-in-automotive.com/aiia/114/yonatangeifmanAI in Automotive Podcast

AI-powered ADAS and AD systems are getting more complex with every iteration. New sensors, higher resolution cameras and increasingly sophisticated deep learning algorithms have substantial computational requirements. Traditionally, they have relied on compute resources in the cloud. But as systems get more advanced, relying on the cloud alone carries significant risk to systems that are supposed to work in real time under significant cost, space and power constraints. So how do we deal with this reality?Enter Edge AI. An elegant solution that helps process part of the input from various sensors locally, rather than in the cloud. In other words, ‘at the edge’. In this episode of the AI in Automotive podcast, we are joined by Orr Danon, CEO of Hailo Technologies, a company that is bringing powerful edge AI solutions to the automotive industry. Hailo’s technology and processor have the capability to process several high-resolution video inputs in real time with low latency, without impacting the accuracy of the algorithm.Orr and I dig deeper into what exactly edge AI is, what advantages it delivers and what its relevance is to ADAS and AD systems. We also talk about the future of autonomous driving, and discuss how the levels of autonomy might evolve in the future. I hope you enjoy our chat today as much as I did recording it with Orr. If you do, do give our podcast a shout on your social media, or share a link with your friends and colleagues.AI in Automotive Podcast

GPS, or the Global Positioning System, is now ubiquitous as a way for us to pinpoint our location anywhere in the world, plot that location on a reference, often a map, and know where we are. Did you know, however, that even some advanced GPS systems can only deliver an accuracy of about 25 cm. This level of precision, while sufficient for you and I, just does not cut it for a vehicle equipped with advanced driver assistance and autonomous driving features.Precise location data is absolutely essential for ADAS and AD functionality. Cameras, radars, LIDARs and other sensors can help the car ‘see’ its environment, but for the vehicle to make sense of the input from these sensors, it needs a reliable, effective way of plotting this world on a reference map.I learned from my guest on the latest episode of the AI in Automotive Podcast that GPS has a number of limitations when it comes to ADAS and AD applications. In this episode, I am joined by Joel Gibson, Executive Vice President of Automotive at Swift Navigation. Formerly, Joel was the Vice President of ADAS, Business Development and Strategy at Magna Electronics, where he started the camera product line and grew that business to be the largest camera automotive tier-1 supplier globally. Joel has 15 patents and holds a BS in Systems Engineering from Oakland University.Joel, in his distinct style, lays out the basics of mainstream location systems, their limitations, and how these are compensated. We then go on a whistle stop tour of Swift Navigation’s technology stack, and how it is making ADAS and AD applications possible with its high-precision positioning service.This episode of the AI in Automotive podcast is slightly different from our usual episodes featuring interesting applications of AI in the automotive and mobility industries. In this episode, we are going a bit further up the data chain and exploring an interesting way in which high-quality data is made available to an AI algorithm.If you find this episode interesting, do share it with your friends and colleagues and rate our podcast wherever you listen to it.http://ai-in-automotive.com/aiia/112/joelgibsonAI in Automotive Podcast

Data is the lifeblood of Artificial Intelligence. Quite simply, the better and richer the quality of data, the more capable the algorithm. Now this applies to both, the training data available to train the algorithm, but importantly, also the input data that is available for the algorithm to do its job.Take the case of autonomous vehicles or advanced driver assistance systems. These systems rely on the eyes - cameras, LIDARs and RADARs - to see the environment around the vehicle. The input from these eyes is then passed on to the brain - the algorithm - which makes sense of what the eyes see. Most state of the art ADAS and AV algorithms today are designed to perceive what these sensors see by drawing bounding boxes around road users. That’s how they perceive pedestrians, other road users, vehicles and obstacles. But human behaviour rarely fits in a box. And human behaviour has a huge impact on how good or not an AV algorithm is. A bounding box alone is not sufficient to really perceive pedestrian behaviour, for instance. Is that pedestrian about to cross the road? How much risk does this road user pose? Is that a vulnerable road user?Enter Humanising Autonomy. A company on a mission to create a global standard for human interaction with automated systems. This is an incredibly interesting company, and I was delighted to have the opportunity to speak to their Co-founder and Chief Product Officer, Leslie Nooteboom.Think of Humanising Autonomy as a module you could add to the AV brain, that then makes the brain capable of perceiving - and predicting - human behaviour on roads. I would imagine a solution like this could improve road safety by orders of magnitude.These guys are up to some really fascinating stuff that sits at the intersection of behavioural psychology, vision perception and artificial intelligence. How does that impact the world of autonomous driving? Find out in my very interesting chat with Leslie.http://ai-in-automotive.com/aiia/111/leslienooteboomAI in Automotive Podcast