Every industry places massive efforts in its product development cycles. It’s no secret that creating an automobile is difficult, but with the rise of software-defined vehicles bringing even more complications to the development process, it’s more important than ever to produce an efficient product development strategy.
With the much faster pace of software development finding its way into the automotive world, it’s imperative to leverage anything to accelerate your product development. Whether it be leveraging a partnership or finding ways to better vertically integrate in your niche, looking for every and any opportunity.
Below, we’ve compiled some examples of product development strategies in the autonomous driving world that can hopefully give you a better understanding of the steps you can take for your project.
Tesla’s Autopilot Full Integration
One of the most publicised product developments in the autonomous driving space is undoubtedly Tesla’s Autopilot. Initially, Tesla entered a partnership with autonomous driving systems provider Mobileye. The full extent of the partnership may not be known. Still, at the very least, it allowed Tesla to see the limitations of autonomous driving technology firsthand and understand what areas were lacking. After several years, the partnership dissolved, but Tesla was able to gather great foundations for building their systems.
One key aspect of creating a system that can navigate almost any situation is how much data you can train it on. Tesla had a significant advantage in this regard. Instead of relying on close-coursed testing like many other autonomous driving providers, Tesla can leverage the data recorded from their hundreds of thousands of vehicles on the road that are driven by ordinary people on ordinary roads. Even if the driver doesn’t have or activate Autopilot, Tesla vehicles have a “shadow mode” that records driving data and weeds out as many corner cases as possible. We, as humans, can use intuition to deal with new situations, but at this point, if a computer system hasn’t seen a scenario, it doesn’t know how to respond.
This influx of constant data significantly accelerates product development as the driving models are constantly updated and then pushed back to the cars through software updates, creating a positive feedback loop.
To increase the speed of product development further, Tesla created a supercomputer known as Tesla Dojo. Using custom-designed chips, the computer processes millions of terabytes of data received from Tesla’s fleet, marking and annotating important details for Tesla’s computer-vision responsible for full self-driving.
Nvidia’s Transformation into an Autonomous Driving Powerhouse
While not an automotive brand, Nvidia has made significant strides in the autonomous driving sector over the past decade. While they were mostly known for their graphics processing units (GPUs) being used for video games, they found a very lucrative niche when automakers started to take self-driving more seriously.
Compared to central processing units (CPUs), GPUs are designed to process data in parallel. Since every autonomous driving suite uses cameras, radars, and LIDAR, all sending data simultaneously to paint the whole picture, GPUs are the best choice to take all this data simultaneously and make split-second decisions.
In 2015, Nvidia introduced a version of the GPUs that were tailor-made for self-driving applications and leveraging neural net technology. As Nvidia is so well known in the semiconductor world, it wasn’t surprising that the company was quickly able to garner dozens of partnerships from technology and automotive partners alike. Amongst these partners are Volvo, Mercedes, and Hyundai.
As with Tesla’s autopilot, gathering data is critical in building a successful autonomous driving system. This is true not only from the software standpoint but also from the hardware side. With so many partners using Nvidia chips, Nvidia can quickly identify how to best build their subsequent chips; while it may not happen as quickly as software updates, it still helps Nvidia remain an industry leader.
These partnerships, however, extend beyond just on-vehicle chips. Nvidia also produces GPUs to be used in conjunction with one another to gather and train data for autonomous driving systems (like the previously mentioned Tesla Dojo). So, even if an automaker creates its in-house chips for its sensor suite, it may still use Nvidia GPUs to process the data and train the software.
While this has helped Nvidia secure its position in the autonomous driving world, it has also helped accelerate the development of its artificial intelligence chips due to the stringent requirements for data processing and decision-making required in driving a car. Some partnerships, such as those with Mercedes, have yielded a more explicit use of AI actually being used in the vehicle’s infotainment system, providing Nvidia with even more data points.
Aurrigo Finding An Untapped Market
But you don’t have to be a multi-billion Pound company to display efficient product development in this space. Aurrigo has spent many years producing various parts for automakers such as Volkswagen and Bentley. The brand ventured into creating autonomous vehicles, but unlike many of the startups we see in the news, Aurrigo went for slow-moving vehicles. These would be cheaper to produce, and the computational requirements would be much less as the system would have more time to react to objects if it’s travelling 10 km/h vs 100.
The company identified an unfulfilled niche in the autonomous driving world as every other automaker tried to create the most desirable vehicles on the road. Instead, Aurrigo looked towards airports and their fleets of baggage tractors. With a virtually untapped market, Aurrigo quickly established partnerships with eager airports such as Singapore’s Changi Airport. These airports allowed Aurrigo to go through trial runs with mock planes and baggage, allowing airport officials to see if such a system is viable and allowing Aurrigo invaluable data to improve their product further. Such public displays occurring at major airports have also allowed the company to gather more investors, allowing it to accelerate its product development further and deploy it worldwide.