Machine Learning for Energy Management in Construction Robots (MACRO)

Volvo material rehandler
Picture courtesy of Volvo Construction Equipment – © 2000 – 2021 Volvo Construction Equipment – All rights reserved

This project develops a methodology to simplify the synthesis of controllers for energy management based on optimal control principles and machine learning. The methodology will be applied on different test cases, different construction machinery, systems, and drive situations, to evaluate the usefulness in different situations. It is a 3-year project supported by Volvo CE and the Swedish Energy Agency (Energimyndigheten) within FFI.

In modern motion systems, components are electronically controlled and generally, a major part of the development time is put into generating control systems to get most performance out of a mechanical or hydraulic system. When it comes to hybrid drivelines the complexity in controls development is an even bigger challenge since there is an additional choice where energy should be taken from and used at. The additional option increases the complexity not only for the engineers when to develop the motion system but also to obtain high energy efficiency while operating the machine in the off-road conditions. The far most common approach to energy management is to apply ad-hoc rule-based strategies where experts need to have deep domain knowledge to derive rules. The effectiveness of any rule-based strategies is highly dependent on the competence level of the engineer and typically a vast amount of time and resources are spent on refining such implementations. Today the energy efficiency is often limited by the skill level of the responsible control engineers.

Addressed Challenges

The research project is addressing three main challenges for develop construction robots and are summarized below:

  1. Reducing the time to market of new machine concepts with improved energy efficiency is a critical aspect to reach the climate change goals. One target in this project is to improve the quality in concept evaluation, reducing the risk of stalling the development to reach the market quicker. The design process and suggested methodology addresses the increased complexity when considering the machine main sub-systems; propulsion and work implement actuation, where actuation is based on hydraulic drives.
  2. Improving energy efficiency is essential for the future of construction machinery. This research project aims at reducing energy losses in the motion systems by improved control principles, allowing concepts to operate closer to their theoretical optimum, more importantly when in real-time operation. To achieve the best possible complete machine efficiency the controls of both propulsion and actuation systems must be coordinated, which is today a significant challenge due to their strong interaction during machine operation, especially in wheel loaders.
  3. Automation is important for future machines. Increased productivity, increased safety and reduced energy consumption are positive outcomes of automation. In autonomous machines operation can be carried out based on pre-information of an assigned work task and machine data at hand. A sophisticated control system can make real-time control decisions with big impact on the total energy consumption. The expected result from this topic in the project will be a reduced electric energy consumption. This will lead to a better utilization of installed battery capacity allowing for improved operating range and/or usage of smaller batteries, reducing the environmental footprint.


Research in Fluid Power