An AI (artificial intelligence)-based system is only as good as its algorithms and the data those algorithms have to work with. When it comes to AVs (autonomous vehicles), the AI must be very good—as good as it could possibly be—to ensure the safety of these systems in an incredibly demanding, dynamic environment. A ResearchandMarkets report that analyzes the global AI in automotive and transportation market says the space will grow at a 20% CAGR (compound annual growth rate) between 2021 and 2026. Growth in this sector during the next five years will be thanks to the increasing popularity and normalization of autonomous vehicles and advancing technologies that continue to make these vehicles as safe as possible in real-world environments.
Pilot programs and services featuring autonomous vehicles are making AVs a more common sight in many cities around the globe. For instance, in September, Ford, Argo AI, and Walmart announced they’re working together to launch an autonomous vehicle delivery service for Walmart customers using Ford self-driving test vehicles and Argo AI’s Self-Driving System. The service will first launch in Miami, Fla., Austin, Texas, and Washington, D.C. Because Walmart is such a large retailer, its use of self-driving vehicles to deliver packages to customers’ homes will put this technology into the paths of many, normalizing it and making it obvious to consumers how valuable it can be in their daily lives.
However, accepting home deliveries from an AV service and buying an AV as a daily driver are two very different things. The 2021 Deloitte Global Automotive Consumer Study suggests many consumers aren’t ready for autonomous features, with only 32% of respondents in the U.S. saying semi-autonomous drive mode is an important advanced feature they’re looking for in their next vehicle. And while safety is one of the reasons AVs will ultimately become the new normal, thanks to their expected ability to reduce accidents that would otherwise occur due to human error, concern over AV safety is also one of the key hurdles facing the space.
A unique new dataset could address this hurdle by greatly improving the algorithms AVs use to interpret and understand their surroundings. The open dataset from Sweden’s Chalmers University of Technology, called Reeds, is now available to researchers and specialists around the world to help boost the development of self-driving vehicles worldwide. Chalmers and its partners say the Reeds dataset was developed using an advanced research boat—a former pilot boat from the Swedish Maritime Admin. equipped with highly advanced cameras, laser scanners, radar, motion sensors, and positioning systems. The boat travels predetermined routes around western Sweden under various weather and light conditions.
The boat’s advanced camera system generates 6 gigabytes of image data per second, resulting in a staggering 16 terabytes of image data per 90-minute trip. The dataset will grow exponentially as the tours continue for the next three years. Researchers say they chose to use a boat because movements relative to its surroundings are more complex than for vehicles on land. By recreating the most challenging conditions possible, the researchers hope the dataset will increase the complexity of software algorithms used in AVs and therefore make them safer.
The open dataset certainly sets a new standard for evaluating the algorithms of autonomous transport systems not only on roads but also in the water and in the air. This important work is a crucial step in helping the self-driving vehicles of tomorrow use artificial perception to interpret and understand their surroundings in a more human-like way. Only then will these vehicles live up to their potential to reduce road accidents and save lives by being smarter and more aware than humans can possibly be behind the wheel.
Want to tweet about this article? Use hashtags #IoT #sustainability #AI #5G #cloud #edge #digitaltransformation #machinelearning #bigdata #AV #autonomousvehicles #artificialintelligence #transportation #mobility