We will not use GenAI in Daimler trucks anytime soon: Raghavendra Vaidya

We have created a bunch of tools to automatically analyse what is going inside the truck and predict which part fails when. That makes our warranty business more profitable, says Vaidya

Daimler AG, the former parent of Mercedes-Benz, split its passenger car and commercial vehicle businesses in 2021. Daimler Truck Innovation Centre India (DTICI) is the engineering arm and technology backbone of the truck and bus business, Daimler Truck AG. RAGHAVENDRA VAIDYA, managing director (MD) and chief executive officer (CEO) of DTICI, explained to Avik Das why the centre in Bengaluru matters. Edited excerpts from an interview at Vaidya’s office:

 

What are the key technology trends in the commercial vehicle industry?

 

The two key trends currently are power to drive and intelligence to drive. The first is the movement to zero emissions; it includes battery electric vehicles, hydrogen gas engines and making trucks with electric power trains. There are huge headwinds in terms of infrastructure, such as producing, transporting, storing green hydrogen or charging stations, and the price parity.

 

Intelligence to drive means the software and electronics to make trucks more intelligent and comfortable, and that is where this centre plays an important role.

 

Can you describe areas where the Bengaluru centre plays a critical role?

 

We have complete ownership and accountability for connected trucks globally. The teams here continuously update the software over the air to make them safer and more intelligent without having to bring those trucks to the service station. The algorithms for emergency braking, such as how to help the trucks stay in lane, prevent lane change when a vehicle comes behind it, stopping the vehicle due to an obstacle and fuel economy, are all done from India.

 

Since these trucks are connected to the Cloud, it makes them vulnerable to cyberattacks. The entire work of keeping them safe is done here by implementing the right technologies. Besides this, this engineering centre also works on battery packs, battery management software and simulating battery conditions.

 

The other area that we are keenly focusing on is data and AI. 

 

We have created a bunch of tools to automatically analyse what is going inside the truck and predict which part fails when. That makes our warranty business more profitable. When we sign long-term agreements, we can tell the fleet owner to come to the workshop and get the parts changed. That reduces warranty cost for us and unplanned downtime for the fleet owner. 

 

Will generative AI (GenAI) be integrated with Daimler trucks?

 

We are not getting GenAI in our vehicles anytime soon. What we use is narrow AI, which includes machine learning (ML) and building models. We use GenAI in our internal operations such as sales, operations, accelerating leads for sales, making contracts more robust, procurement and production. How can you use a system that is predictive in nature and can hallucinate to control trucks? We use foundational ML models and narrow AI techniques to make our trucks better. For example, many of our advanced driver assistance systems algorithms are predictive in nature but that is not GenAI. We use AI techniques to predict the range of electric trucks or physics-based models based on weather, terrain, and past data driver behaviour. Such models can be used to solve a small problem, such as a lane departure warning or predicting the failure of parts in the trucks.

 

Where can GenAI make an impact at Daimler Truck?

 

We generate codes that are monitored by human programmers. You cannot take the output of GenAI and have it make machine critical decisions. When we get to agentic AI, we will have agents that are smart, reliable and predictable, but we are not there yet and certainly not with our product. We can use agentic in company operations, for example, how do we give tools to our after-sales people. If the after-sales person is sitting with a customer and the customer is negotiating for a certain type of bus, the agentic AI can tell me that there is a model very similar to the needs of the customer and I can pitch it. We have to understand what is the AI paradigm we use and that goes into the product, as it has to be bullet proof.