• Ansprechperson:

    Prof. Dr.-Ing. Marcus Geimer

  • Förderung:

    European Union’s Horizon 2020 research and innovation programme

  • Projektbeteiligte:

    Academic Beneficiaries

    - Karlsruhe Institute of Technology, Karlsruhe, Germany

    -Tampere University, Tampere, Finland

    -Örebro University, Örebro, Sweden

    Industry Beneficiaries

    -Novatron Oy, Tampere, Finland

    -Liebherr, Bischofshofen, Austria

    -Hiab AB (Cargotec), Hudiksvall, Sweden

    -John Deere Forestry, Tampere, Finland

    -Bosch Rexroth AG, Elchingen, Germany

    Partner Organisations

    -Volvo CE, Hallsberg, Sweden,

    -Technische Universität Darmstadt, Darmstadt, Germany

    -GIM Oy, Espoo, Finland

    -accelopment Schweiz AG, Zürich, Switzerland

  • Starttermin:


  • Endtermin:


Research topic:

Work performance evaluation of mobile machines in earth moving tasks by using process models and sensor data

Link: https://www.more-itn.eu/early-stage-researchers/amirmasoud_molaei/

Project description

MORE is an innovative European Industrial Doctorate (EID) research and training programme. MORE will deliver innovative solutions driven by digitalization and Artificial Intelligence (AI) in three areas:

Processes: environment modelling by combining machine-earth interaction with vision, and optimizing material flow in construction sites, business processes.

Machines: creating efficient power-train solutions for heavy-duty booms and

Control: object classification and environment modelling to support long-term autonomy; obstacle avoidance in adverse conditions; innovative solutions on transfer-learning for earth moving and boom control to reduce operational and development costs, in different application areas.

Project tasks and objectives:

In order to evaluate the performance of a mobile machine, various data has to be considered, since mobile machines do have travel and working functions. For example, when only looking at the overall fuel consumption of an excavator, the fact is neglected, that the following steps in a process chain also depend on the timely fulfilment of the designated task.

For autonomous machines to work efficient, it is important that the machines are able to evaluate their own working process and its performance level, according to specified parameters. Based on the evaluation result, the machine must derive and implement necessary changes in order to optimize its working process. Therefore, being able to process data, organize knowledge, evaluate the performance, make decisions based on results, act accordingly and learn from experience are prerequisites for intelligent and autonomous actions and systems.

The work includes:

Research of State of the Art concerning earthmoving processes, autonomous mobile machines, process evaluation and work cycles.

Research and studying of applicable modelling strategies for process evaluation of a mobile machine in an earth moving process.

Investigate earth moving processes and interview machine operators in order to gather data for the performance definition.

Define a suitable performance vector for the target application, e.g. by using work efficiency in metric tons / hour and fuel consumption for both digging and dumping processes.

Devise sensor concepts for an autonomous machine to acquire the necessary data during real-time operation.

Develop the required data processing software by using both model-based computational and model-free machine learning techniques.

Evaluate the performance of operations in the context of the target landscape given by BIM infra models.

Define and evaluate reward functions for reinforcement learning.

Verify / validate functionality in simulation.

Derive a method from the results to evaluate a process of a mobile machine and get a quantitative measure of its system performance, which can be used in further machine-learning purposes and improved autonomous operations and to select the most efficient working method and tool for a given task.