MPC allows heavy-duty vehicles to drive by themselves

Published Feb 11, 2016

Autonomous vehicles have become a popular research field in the last decade. Mostly, the research is conducted with private cars or other small vehicles in mind. But where does that leave the industry with their larger vehicles such as trucks? Pedro Lima, a PhD-student at the Department of Automatic Control, has designed and implemented a Model Predictive Control (MPC) for trucks that will allow them to drive safely and accurately on their own.

During the last two years, Pedro Lima has been researching a control design called Model Predictive Control (MPC), a control method that is very popular at the moment. MPC allows vehicles to drive by themselves with no human involvement. Besides the wow-factor of having driverless cars in society, there are several major benefits that come from using Model Predictive Control..

“First, as the name indicates, MPC uses a vehicle model to predict the motion of the vehicle for a given set of inputs, such as steering angle, gas and brake, for a certain time-window. We can minimize the deviation from the path, maximize the comfort of the passengers by minimizing lateral jerk and acceleration and maximize the efficiency of the vehicle by having a model of how much fuel the vehicle consumes", says Lima. "When designing the controller one can choose what to benefit. Also, it is a very systematic way of designing the controller and posing the problem and it handles non-linear time-varying models and constraints, which in our case is necessary and very useful."

The difference between Lima’s research and others’ within the same area is the focus on autonomous trucks. Never before has MPC been applied in heavy-duty vehicles. The goal of the project is to make it possible for industrial companies to automate entire parts of their operation, for example in underground mines or in construction. This would substantially minimize the risk of human fatalities. At the same time, it will increase productivity and efficiency in these working sites. 

“Heavy-duty vehicles differ significantly from normal passenger cars due to their enormous mass and inertia. Therefore the vehicle dynamics involved are much more prominent. These result in a more complex model, which would be very difficult to use in an MPC. In the case of autonomous driving, the vehicle controller needs to run in real-time. Typically, we need to compute a new input (gas, brake, steering) every 50ms. Therefore, very complex vehicle models are also very time consuming” says Lima. 

In collaboration with Scania

The research is conducted within iQMatic, a project led by Scania where the Department of Automatic Control is one of the involved parties. The goal of the project is to develop fully autonomous truck for mining operations by 2018. In December last year, the MPC controller was tested at a demo track.

“The path was completed with an average deviation from the centerline of the road of 10-20cm and with no human intervention. Besides being very accurate, the controller was also able to drive at high-speeds. The maximum speed for trucks is 90 km/h and the controller handled that in a smooth and stable way”, says Pedro Lima.

The project is in the middle of its lifetime. In May, there will be another demo where the truck will be fully autonomous. This means, that no human intervention is necessary and the system is able to react and reason in real time in (almost) every case.

“The demo will consist of a realistic mining scenario, where two trucks should cooperate autonomously to perform several orders to load and unload gravel", says Lima. "Our participation in the demo will contribute with the controller aforementioned. Later this year, we intend to test the vehicle in an underground mine in the north of Sweden. That brings a whole new set of challenges, from the absence of GPS signal to the extremely narrow and difficult roads.”

Autonomous vehicles – not in a faraway future

In the future, he believes that fully autonomous vehicles will be a reality in less than a decade. This is indicated by the fact that the main automotive manufacturers are investing large amounts of money and resources in developing the first autonomous vehicle to be mass-produced.

“The development of autonomous vehicles will bring enormous benefits to the world population. Road traffic injuries are one of the main causes of death worldwide; in major cities, money and time is wasted in traffic jams; greenhouse emissions need to be significantly reduced in the next decades. Autonomous vehicles can help solving all the aforementioned” Lima concludes.

Contact

For more information on the project, contact:

Pedro Russo De Almeida Lima
Pedro Russo De Almeida Lima, ph.d. student
Top page top