July 14, 2024

Was the challenge of autonomous driving underestimated by carmakers?

Recent research conducted by the US’ Insurance Institute for Highway Safety (IIHS) has revealed that car manufacturers need to improve their partial driving automation systems, raising doubts about the feasibility of fully autonomous driving in the near future.

As part of a new ratings program introduced by the IIHS to incentivize automakers to enhance the safety measures in their automation systems, 14 systems were initially tested. Among them, only one received an acceptable rating, while two were rated marginal and 11 poor.

IIHS President David Harkey commented on the systems tested, stating, “Most of them lack sufficient measures to prevent misuse and ensure drivers remain attentive to the road conditions.”

The results of the study raise concerns for a market that GlobalData’s Automotive ADAS and Autonomous Vehicles Market report estimates to have included 516.3 million units in 2023 and predicts a compound growth rate of over 6% between 2023 and 2028.

These findings contrast with the vision of a fully autonomous future promoted to consumers by car manufacturers in recent years, particularly by Tesla over the past decade. Even the Tesla Model 3, one of the vehicles tested, received poor ratings for its partial automation systems.

However, the IIHS study is not the only indicator suggesting that achieving fully autonomous driving may take longer than initially anticipated. General Motors’ Cruise self-driving subsidiary laid off 900 employees in December after safety concerns led to the removal of its vehicles from the roads. Additionally, reports in February indicated that Apple had abandoned its highly anticipated car project, which was believed to prioritize full autonomy.

Dr. Ilja Radusch, head of the Daimler Center for Automotive IT Innovations at TU Berlin, highlighted that the challenge of achieving fully autonomous driving has been underestimated. Following the Cruise incidents in February, he expressed, “It’s disheartening to realize that we still need to consider what was deemed remotely possible, entailing a significant amount of additional work.”

The extent of work required is a question posed by experts within the industry, as explored below.

Sammy Chan, manager for automotive sales forecasts, GlobalData (business intelligence)

Several car manufacturers and other ventures have reduced or scaled back their investments in autonomous driving in recent years, likely due to underestimating the complexity and timeline required to achieve fully autonomous capabilities.

Ford and Volkswagen initially invested in Argo.AI to pursue level 4 capabilities but withdrew from this joint venture in 2022 to focus on level 2 and 3 offerings instead. The companies found that development costs were escalating, and the commercial returns from level 4 would still be distant in the foreseeable future.

Challenges also include navigating the regulatory landscape and building public trust. Accumulating driverless miles for data is crucial for companies striving for fully autonomous driving. However, even industry leaders face significant setbacks. Cruise is facing challenges to regain its driverless car permit in California following an incident in October 2023 where a Cruise vehicle collided with a pedestrian.

Dave Kelly, chief corporate officer, Cubic (connected vehicle software)

This recent study adds to the mounting evidence indicating that automotive OEMs are struggling to meet stringent standards concerning driver safety. OEMs have primarily focused on enhancing the in-car driver experience through improvements in connectivity, local content, and user interfaces. While these features are crucial with the increasing adoption of autonomous driving, a key aspect driving the autonomous vehicle market is the promise of a safer journey for drivers, passengers, and pedestrians. Regulators are grappling with reconciling technological advancements with ensuring road safety and are introducing new regulations to prevent drivers from getting dangerously distracted.

For instance, starting July 2024, the EU will mandate that all new vehicles be equipped with driver drowsiness and attention warning systems. These systems will detect driving and steering patterns indicative of reduced alertness or attention, ensuring drivers using partial driving automation systems remain focused on the road.

For automakers aiming to deliver software-defined connected services for achieving level 4 and 5 vehicle autonomy, it is critical to align the rollout of self-driving car technology with these regulatory developments from policymakers. This alignment presents a challenging hurdle in the quest to mainstream and commercially viable autonomous vehicles.

Kia Cammaerts, technical director, Ansible Motion (driving simulators)

The increasing need for a safe, repeatable, and cost-effective method of capturing and analyzing human behavior is leading more OEMs to fully embrace driver-in-the-loop simulation for advanced driver-assistance system (ADAS) validation and testing.

While autonomous systems may initially appear to render human drivers redundant in developing a new vehicle, none of these vehicles will operate independently. Their behavior must be safe, comfortable, and instill confidence in their human occupants. With the rise of connectivity, electrification, and the diverse scenarios requiring human driver input, it could take thousands of years to validate nearly every potential scenario.

To ascertain how ADAS interventions will impact driving pleasure, OEMs and tier-one vendors need to understand how humans interact with these systems. Does the vehicle’s behavior inspire confidence? Is it comfortable? Does it reflect the brand’s essence? Using a real person in an engineering-class driver-in-the-loop simulator can expedite finding answers to these questions.

Dennis Marcus, Commercial Manager, Cruden (driving simulators)

One of the challenges observed from the less-than-successful implementation of certain ADAS in current vehicles is the lack of acceptance by drivers. These systems were developed and tested by engineers without involving real people or utilizing driving simulators.

However, some manufacturers have established simulator test centers, such as BMW in Munich and planned by Porsche, where hundreds of members of the public are invited to participate in the development process to evaluate different automotive systems.

This type of simulator test center is poised to result in better vehicles. OEMs need to understand consumer preferences. Incorporating more real people in simulated testing will reveal human preferences and lead to the development of more acceptable systems, especially human-machine interface strategies, which have proven unacceptable to customers when insufficiently tested. For instance, the recent shift from all-touch screens back to buttons in cars indicates the importance of understanding human preferences through widespread simulator usage.

John Ellmore, editor, Electric Car Guide (media outlet)

Following the recent findings by the Insurance Institute for Highway Safety (IIHS) highlighting the shortcomings in the first wave of partial driving automation systems in terms of safety protocols, it appears that carmakers may have underestimated the complexities and challenges associated with autonomous driving.

The evaluation underscores the significant room for improvement in ensuring the safe operation of these systems. Striking a balance between technological advancement and ensuring public safety on the roads, continual innovation and rigorous testing are essential before fully autonomous vehicles can become commonplace.

Richard Lawton, head of marketing and communications, DriveElectric (EV leasing and charging)

At DriveElectric, we acknowledge the complexity involved in developing hardware and software systems for autonomous driving and do not believe there has been an underestimation of these challenges. While the development journey has been non-linear, significant strides have been made in enhancing the safety of cars year by year and model generation by model generation.

The technology and computational capabilities of the latest electric vehicles were unimaginable just five years ago. While carmakers face hurdles to overcome, the advancement of increasingly sophisticated AI platforms deployed to assist in developing autonomous driving models indicates that the first fully autonomous vehicles may not be too distant in the future.

Leave a Reply

Your email address will not be published. Required fields are marked *