Bauer Completes AI-Driven Control Platform Project for Foundation Engineering

The OPAI4DNCS project—jointly developed by Bauer, Hawe Hydraulik, STW, and two chairs from the Technical University of Munich—focused on creating operator-centered assistance systems using artificial intelligence. At the heart of the project is a shared-control platform designed to enable intuitive collaboration between human operators and AI systems, successfully demonstrated on a BAUER GB 50 hydraulic grab.

Operating large, mobile hydraulic construction equipment is a demanding task. Safe and efficient operation depends primarily on the operator’s experience – expertise that can take years to acquire. Less experienced operators often reach their limits quickly. This is precisely where the recently concluded OPAI4DNCS research project, funded by the Bavarian Research Foundation (BFS), stepped in.

Industry and Science Joining Forces

For this project, Bauer, HAWE Hydraulik SE and STW (Sensor-Technik Wiedemann GmbH) collaborated closely with the Chair of Automation and Information Systems as well as the Chair of Ergonomics at Munich Technical University. Over three years, the project partners explored operator-centered assistance for mobile construction equipment used in specialist foundation engineering.

Needs-Based Assistance from AI

This project aimed was to develop adaptive, intelligent, and self-learning control systems based on multi-agent systems (MAS). The focus was on needs-based assistance of machine operators to enable safe operation even in borderline situations.

Bauer Grab Used as a Test Vehicle

The technology was demonstrated using a hydraulic grab, typically used in the construction of diaphragm walls. The crucial factors were dampening vibrations and stabilizing the grab. Operating the grab is particularly challenging due to its swinging bucket. Using a soft sensor to measure the pendulum movements caused by pressure fluctuations on the hydraulic actuators, a real assistance system was development and successfully implemented in test campaigns.

Human-AI ollaboration

One central result of the project was the development of a shared-control platform that enables intuitive collaboration between human operators and artificial intelligence. A retrofitting solution was established that can extend existing control systems with intelligent functions, without requiring major changes to the machine’s architecture. The assistance system offers three different control modes tailored to the operator’s experience level. Combined with a multi-agent architecture, the system delivers flexible, context-sensitive assistance for complex control tasks – improving efficiency, safety and resource conservation on construction sites.