Within a decade, a quarter of all cars will be capable of Level 4 or 5 autonomy, believes Mindtree. Megan Lampinen talked to the company to learn more about an auto industry driven not by drivers, but by data
In terms of technology, much can happen in a decade. Consider, for example, the changes in consumer electronics in the decade since the introduction of the iPhone. Across the automotive industry, an unprecedented rate of change is revolutionising mobility concepts, and automated driving in some form or other is generally expected to pervade the global transport industry within the next ten years. While this should entail massive safety and emissions improvements, the road towards that goal is riddled with potholes.
“Our vision is that all cars will have some level of automation, from Level 1 to Level 5, by 2026. About 25-30% of cars will have full autonomy at Level 4 and 5, meaning in some cases there is no driver, no steering wheel.” That’s the view of Murali JP, Senior Director and Automotive Business Leader at Mindtree.
The company, headquartered in Bangalore and New Jersey, operates in five key connected vehicle areas – infotainment, telematics, navigation, safety systems and diagnostics – and collaborates with global OEMs and suppliers on technology to support autonomous vehicles. Notably, its client list also includes aftermarket system manufacturers, and it believes these will play an important role in autonomous mobility. “In the next ten years, every car will be connected, including aftermarket cars,” said JP. “The entire ecosystem will transform into the connected and autonomous space.”
Vehicle manufacturers and Tier 1s are putting safety ratings and standards compliance first and foremost. Many of those things continue to evolve. That is the journey that our customers are taking
Predicting monetisation aspects in this autonomous ecosystem is a guessing game at the moment, but Mindtree expects the traditional OEMs to continue playing a starring role in the set-up. After all, the industry may transition to an autonomous car-sharing scenario but someone still has to provide the cars. “Asset ownership will still lie with automotive OEMs,” JP suggested. “There is a revenue stream that they will tap into.”
That said, he also expects the technology players, including some of the Tier 1s, to grab a larger slice of the pie moving forward. This is due to the rising software content in vehicles as well as the system integration which will be required by the many vehicle variants. “System integration is a huge task, and some of the Tier 1s will be active in that,” noted JP.
Filling the gaps
This transition to a fully connected, highly autonomous future will come in phases, albeit quickly moving ones. The industry is already off and running, but there are some substantial holes that need to be plugged before such a transition can be completed.
“The transition is imminent,” said JP. “With the shift towards connected cars, we see major challenges in terms of the explosion of data that is going to happen. Handling this data will be a major issue. Who will support the data transfer between the car and the back-end for the analytics and insights into customer behaviour? Then there is the information that needs to be pushed in as part of the content into the car. Connectivity will be a big issue from the use case perspective and even from the safety angle.”
One significant concern stems from interruptions in connectivity. At present, in many markets, connectivity is not necessarily guaranteed everywhere a car may travel. “That means some of the things for safety could be done with edge computing, but not all,” he warned. “There will be numerous safety cases that need to be addressed.” Along with these technology holes come cyber security and privacy concerns.
Our vision is that all cars will have some level of automation, from Level 1 to Level 5, by 2026. About 25-30% of cars will have full autonomy at Level 4 and 5, meaning in some cases there is no driver, no steering wheel
For Prasanna Gopal, Senior Director in Mindtree’s Solutions in Networking & Communications, Semicon and Connected Vehicle Segments, legal issues are a big grey area. “The most hotly debated topic at the moment is where responsibility lies when a self-driving car has an accident. That is a legal point where governments will need to come forward and work with the industry,” he told Megatrends. Then there are the regulation aspects and the official standards, many of which are yet to be finalised. “The industry is going ahead with the implementation of autonomous technology, but there are open items in many areas that still need to be standardised,” cautioned Gopal.
The execution of the connected car itself will play an important role in realising a successful autonomous future, and this includes pricing strategies for automated features. “When these automated features come to market, they will carry a price premium. As things move forward, people will want to see lower costs for wider adoption. Once you have this wider adoption, then you will see the benefits of the autonomous car,” Gopal explained. This, he noted, is particularly true for vehicle-to-everything technology.
Evaluating the technology
When it comes to evaluating autonomous technology, OEMs and suppliers will need to pursue a mix of simulation and real world testing. The number of vehicle manufacturers and technology companies granted road testing permits is steadily growing, with new regions passing legislation to support on-road pilots on an almost regular basis. At the time of writing, California alone has issued autonomous vehicle testing permits to 36 companies. However, this approach is much more expensive than simulated evaluations. The key will be in achieving the right balance.
“For verification, companies need to simulate scenarios as much as possible during the development and integration phases,” recommended Gopal. “It is very expensive to do multiple rounds of road testing.” As much as possible, he said, they try to complete those rounds of testing through simulation. “That is currently happening and it will continue to happen.” At a later stage, real world testing is essential and, Gopal added, unavoidable.
With the shift towards connected cars, we see major challenges in terms of the explosion of data that is going to happen. Handling this data will be a major issue. Who will support the data transfer between the car and the back-end for the analytics and insights into customer behaviour?
JP shared a similar view. “When we speak to industry players, we keep hearing that simulation is good for all the base use cases. Simulation can realistically create the scenarios to gauge the car’s response for around 65% to 70% of the time. The remaining 30% has to be done in a real world scenario.”
Notably, gathering all the necessary data for these vehicles cannot be done by one party alone. JP predicts that consortia will be required, and already the announcements are flooding in. BMW is working on autonomous technology with a group consisting of Intel, Mobileye and Delphi. The UK Autodrive project, which kicked off in late 2015, is conducting on-road trials using cars provided by project partners Ford, Jaguar Land Rover and Tata Motors European Technical Centre. In fact, almost all major interested parties are partnering up.
Giving it the green light
After all the simulations and the road testing, at some point an autonomous vehicle will need to be deemed safe enough for public use. “It is very hard to quantify what constitutes ‘safe enough’,” observed JP. “It is very subjective to the end customer and the early adopters to begin with.”
Some industry players have suggested various ways of measuring safety, such as the number of disengagements per number of miles. But the questions keep coming, observed JP: “Is it disengaging once every 50,000 miles? If so, what is the driving condition – is it in an urban environment or on the highway? There could be a measure, although there is no number attached to it at this point in time.”
When we speak to industry players, we keep hearing that simulation is good for all the base use cases. Simulation can realistically create the scenarios to gauge the car’s response for around 65% to 70% of the time. The remaining 30% has to be done in a real world scenario
In general, Mindtree backs this approach as a feasible one from a customer acceptance perspective. “In any given geographical area, driving the car around with a fewer number of disengagements per so many miles could be a good measure to quantify that this is safe enough to go. In the end, though, the person who will decide is the user,” JP commented. This, he suggested, would entail more user surveys to fully understand consumers’ views.
“Vehicle manufacturers and Tier 1s are putting safety ratings and standards compliance first and foremost,” said Gopal. “Many of those things continue to evolve. That is the journey that our customers are taking.”
Concluding, JP confirmed that Mindtree’s Connected Car Center of Excellence is working with “all the leading Tier 1s, global vehicle manufacturers and semiconductor suppliers.” Again, the partnership aspect was flagged as pivotal to the industry’s future development, with JP adding: “There is a confluence in terms of partnerships as the players work together to solve some of the critical challenges that connected and autonomous cars face through this disruption.”
This article appeared in the Q3 2017 issue of Automotive Megatrends Magazine. Follow this link to download the full issue