Artificial Intelligence Being Schooled for Mining Applications

Helm.ai looks to transition its AI automated driving for passenger cars to off-highway vehicles.

February 14, 2025

Kami Buchholz

Helm.ai’s perception software system enables autonomous capabilities in mining applications. (Helm.ai)

A battery-electric Honda midsize SUV entering production in early 2026 will use Helm.ai’s artificial intelligence to facilitate conditional automated driving. The start-up firm’s AI technology could soon see its first off-highway application.

“Different driving environments look pretty much the same from an engineering perspective, so the lessons we’ve learned from [passenger vehicle] autonomous driving can be brought to the mining space in a fairly seamless fashion,” Vladislav Voroninski, co-founder and CEO of Helm.ai, said in an interview with SAE Media.

The first production-vehicle application of Helm.ai technology will be Honda’s new 0 Series. Featuring a newly developed EV-dedicated architecture, the 0 Series launches with the 0 SUV and the 0 Saloon. Prototypes of both vehicles globally debuted at CES 2025 in Las Vegas. The SUV will be the first 0 Series vehicle to enter production – slated for the first half of 2026 – with assembly at Honda’s EV Hub in Ohio. Honda’s 0 SUV and 0 Saloon initially will be sold in the North American market, followed by Japan, Europe and other global markets.

Vladislav Voroninski, co-founder and CEO of Helm.ai, spoke with SAE Media. (Helm.ai)

“For the 0 Series, we will combine the wisdom of Helm.ai with our own technology,” Steve Frey, vice president of development operations at the North American Automotive Development Center, Honda Development & Manufacturing of America, said during Honda’s press briefing at CES 2025.

In 2021, Honda became the first automaker with Level 3 automated driving under limited conditions with the launch in Japan of the Honda Legend equipped with Sensing Elite. With 0 Series models, the automaker will expand its use of AI technology globally.

“To do so, we need AI that can recognize and understand its surroundings on its own, in each region which has its own unique traffic conditions,” Frey said, referencing Helm.ai’s Deep Teaching AI technology.

Deep Teaching AI

Since 2016, researchers and engineers at Redwood City, California-headquartered Helm.ai have leveraged the firm’s proprietary Deep Teaching technology to support the continued development of AI for autonomous driving.

“Our proprietary Deep Teaching technology is geared toward unsupervised machine learning using applied mathematical modeling techniques. That allows us to train on large amounts of data with very limited human annotation,” Voroninski explained.

Helm.ai’s Deep Teaching has enabled the creation of scalable pipelines for generative AI simulation as well as the continued development of computer vision, also referred to as a perception software system.

“Computer vision has flexibility advantages in terms of being able to address various corner cases that you might not be able to address with other sensors,” Voroninski said. The ability to detect and react appropriately to an unexpected driving situation underscores why computer vision is a highly useful AI tool.

“Of course, there are other technology pieces that need to be implemented. But once you build a perception software stack, that addresses a very big chunk of the autonomous driving problem that answers the basic questions of ‘Where am I? What am I looking at?’ and ‘What are all the other things around me,’” Voroninski said.

Managing mining hazards

The first production-vehicle application of Helm.ai technology will be Honda’s new 0 Series models, which will be equipped with a system that enables expansion for the range of driving conditions where driver assistance and Level 3 automated driving are available. (Honda)

In a mining environment, possible hazards might lurk underground, or obstacles could be concealed in pools of water or hidden in rocky crevices. “There’s a baseline level of danger associated with mining jobs, and AI systems can help alleviate some of those dangers by either warning humans or by actually allowing humans to execute the tasks in an AI-driven fashion,” Voroninski said.

Real-world field data has been crucial to Helm.ai’s ongoing work on autonomous mining. “We’re getting access to data from some of our partnerships,” Voroninski said, noting that the names of those partners remain confidential for now. “There is definitely a high level of interest in AI solutions from mining OEMs,” he added.

Autonomous mining vehicles could report for field duty in a matter of months. “Turns out our technologies for training computer systems for self-driving cars carryover in a very straightforward way to other domains, like mining,” Voroninski said. “We’ve made a lot of progress in the mining space just in the last year.”

According to Voroninski, “The work that we’ve done to get our autonomous-driving perception software system to production level – both in terms of the accuracy of the neural networks as well as the various types of safety certifications – those experiences and processes allow us to streamline our development for other areas. And that gives us an advantage of several years if you compare it to starting from scratch and just doing mining applications.”