Master Thesis Student

Location: Göteborg, SE, 417 15

Master Thesis Student

Location

Göteborg, SE, 417 15

#job-location.job-location-inline {

display: inline;

}

Position Type

Student

Permanent/Temporary

Temporary

Transport is at the core of modern society. Imagine using your expertise to shape sustainable transport and infrastructure solutions for the future. If you seek to make a difference on a global scale, working with next-gen technologies and the sharpest collaborative teams, then we could be a perfect match.

Thesis Background

Sustainability including climate change are the challenges of our generation. Our contribution is to offer leading transport and infrastructure solutions enabling societies to prosper in a sustainable way. At Volvo group, we are committed to the ambitions and climate change goals of the Paris Agreement. From a lifecycle perspective, most of the emissions occur during the use phase of our products.

Therefore, our priority is to develop solutions that reduce the carbon emissions from transportation. Today, Volvo is the market leader in offering a full range of electric trucks.

We are Vehicle Energy Management team at Trucks Technology and Industrial (TTI).

As the team’s name suggests, we work in the area of energy management, where we optimize the vehicle energy using the predicted information such as topography,curvature, and speed limits from the road ahead of the vehicle. Our goal is to deliver a platform solution to optimize the energy consumption for all three powertrain variants in Volvo (ICE, Fuel Cell and BEV (Battery Electric Vehicles).

Motivation

For electric trucks to match or exceed the utility of conventional fossil-fuel vehicles, intelligent trip planning is necessary. One of the key limitations of electric vehicles (EVs) is the comparatively long charging time of lithium-ion batteries, which reduces vehicle uptime and extends mission duration.

Additionally, the limited availability of charging stations further contributes to range anxiety, where drivers fear running out of battery before reaching a station. Effective trip planning can significantly mitigate this issue. However, optimizing where and when to charge, while accounting for uncertainties such as weather conditions, queue times at charging stations, and fluctuating electricity prices, adds complexity to the problem. In addition, legal regulations mandate regular uninterrupted breaks for drivers. It is therefore crucial to integrate rest periods into the trip plan and align them with charging opportunities.

Description of Thesis Work

The objective of this thesis project is to develop and evaluate a closed-loop, stochastic mission planner for long range electric truck vehicles. The planner decides where to stop, which charger to use, how much to charge, and when to take the mandatory driver breaks. The primary goal is to minimize the total mission duration and cost (travel + charging + driver rest time) while keeping the risk of running low on battery below a specified level. This is done by finding an optimal sequence of charging locations and driver rest stops while accounting for real world uncertainties.

Previous work. This project builds on ongoing research at Volvo and Chalmers

[2, 1]. In the most recent master thesis [2], a charge planner based on approximate dynamic programming was developed. Traffic and wind disturbances were modeled as Gaussian processes and propagated along the route with an unscented Kalman filter. The results were promising, but they were obtained in simulation only and with hand-tuned uncertainty models.

Several problems remain open

The uncertainty models are not grounded in real data.

Information gathered during the mission is not used to update the predictions;

Driver rest periods are not included;

It has not yet been shown under which conditions a stochastic planner outperforms a well-tuned deterministic planner with safety margins.

Approach - The planning problem will be formulated in a Dynamic Programming (DP) framework, which provides a common foundation for Model Predictive Control (MPC) and Reinforcement Learning (RL). Decisions are made at charging stations, and the state includes the battery state of charge, the time of day and the driving time since the last break. Instead of computing a fixed plan, the algorithm will execute a closed-loop policy: at each stop it selects the decision that minimises the expected cost while keeping the risk of a low battery level acceptable, taking into account that later decisions will also adapt to new measurements. The plan is recomputed in a receding-horizon (MPC) fashion as new information arrives, and the expected remaining cost of the mission is learned offline (RL) to keep online computation fast.

Data-driven uncertainty models. Using logged data from Volvo’s electric truck fleet, the project will quantify which sources of uncertainty dominate the energy consumption and travel time of long-haul missions. Live traffic and weather forecasts will provide the expected conditions along the route. The forecast errors will be modelled probabilistically, e.g., with Gaussian processes whose properties depend on road type, time of day and forecast horizon. During the mission, Kalman filtering will be used to update the models with the truck’s own measurements, for example to estimate how the actual energy consumption deviates from the prediction.

The project includes

Analyzing logged fleet data to quantify and model the main sources of uncertainty in energy consumption, travel time and charging, and combining these models with real-time traffic and weather forecasts.

Refining the vehicle, battery and charging models to predict energy consumption and arrival times accurately, together with their uncertainty.

Formulating the mission planning problem, including charging decisions and driver rest periods, as a closed-loop stochastic optimal control problem, and solving it with MPC using a value function learn