"Eyes" for the autonomous vehicle of the future
A necessary prerequisite for autonomous driving is that the vehicle recognizes its surroundings completely and precisely. This is the only way it can "understand" the traffic situation and choose the optimal, safe route. However, vehicles’ fields of vision are limited, which is particularly critical when traffic is highly automated and in complex terrain and driving situations. Two exhibits from TU Ilmenau at the Hanover trade fair present solutions to this challenge by giving the autonomous vehicle of the future "eyes": A drone that serves as an external sensor platform and assistant for vehicles on rough terrain, and AI-supported perception software for autonomous robots moving on sidewalksIf autonomous vehicles are traveling off-road, for example on construction sites, in agriculture or in completely unknown terrain, it is particularly important for them to receive precise information about the terrain situation in good time: Where are there obstacles? What is the road surface like? Is the road surface muddy or icy? To determine all these parameters on the road ahead of the vehicle, a research team from the Systems and Software Engineering Group and Control Engineering Group has developed a patented drone-based method.
A drone equipped with lidar sensors and a computer flies in front of the vehicle at a constant height and detects obstacles in the road, measures the road profile and transmits this information to the vehicle in real time. This enables the vehicle to avoid obstacles in good time, adapt its suspension or anti-lock braking system to the changing road conditions and brake or accelerate accordingly.
The process was developed at the Department of Computer Science and Automation at TU Ilmenau by a research team from the Smart Vehicle Systems working group led by Viktar and Aleksandra Beliautsou: "The drone could be used, for example, as an assistant for rescue or disaster control vehicles. Depending on the terrain and scope of the mission, it is possible to vary the payload of the drone with additional cameras, sensors or measuring devices, to use mathematical modeling to calculate the necessary battery life and size of the drone or to use several drones at the same time."
A patented solution for the autonomous driving of the future will also be presented at the Hannover Messe: AI-supported perception software for autonomous robots traveling on sidewalks. The software, which is based on an algorithm developed in-house, acts as a central perception and control system that precisely records the traffic situation and the environment, enabling the robot to behave both proactively and reactively.
Qais Yousef, who developed the new method together with Prof. Pu Li, head of the Process Optimization Group , explains the major advantage over comparable systems: "While these only use pattern recognition to analyse the pedestrians’ movements, our system also reads their facial expressions and can therefore predict the pedestrians’ intentions in real time. This allows the robot to change its route at an early stage without having to brake at short notice."
Another new feature is the software’s ability to recognize the environment at an early stage and control the robot accordingly: "We give the robot eyes: with the help of a 2D camera, our software can analyze and understand the entire environment - the condition of the sidewalk, the weather situation or different lighting conditions - and react accordingly - even from a distance thanks to digital zoom."
The software is also already prepared to communicate with traffic lights and make statements about the degree of certainty of the AI’s predictions. This ensures safe and efficient navigation between pedestrians and enables the robot to complete its tasks faster and on optimized routes. The software can be used for delivery robots, robots for cleaning sidewalks or assistance robots for visually impaired people, for example.
Powerful and energy-efficient mobile networks for industry
A key prerequisite for autonomous, connected driving, but also for automated and efficient production processes, is a high-performance, stable mobile network. This is why TU Ilmenau is also presenting pioneering innovations for energy-efficient and intelligent private mobile networks, so-called campus networks, at this year’s Hannover Messe. Tailored to the individual needs of users, they will play a central role in the introduction of the future 6G mobile communications standard.Scientists working with Prof. Andreas Mitschele-Thiel, head of the Group for Integrated Communication Systems , have designed a 5G+ campus network as part of the 6GCI project, which serves as a research platform for autonomous systems, industrial automation and augmented reality applications. Intelligent network technologies such as Intent-Based Networking enable intuitive and automated network control. Network operators can formulate their requirements as "intents", i.e. in natural language. The system automatically implements these, monitors performance and adapts itself. In combination with the open radio access network O-RAN (Open Radio Access Networks), which promotes manufacturer-independent, standardized network architectures and will play a key role in future 6G networks, operators can seamlessly integrate new mobile radio technologies without being slowed down by increasing system complexity. This enables powerful and flexible networks for a wide range of industrial applications.
At the same time, the energy consumption of the networks has also been optimized with the help of the Energy Saving (ES)-xApp/rApp. Radio access networks connect mobile devices such as smartphones, cars and networked devices in industry with the core network and account for over two thirds of the energy consumption of mobile base stations. The innovative software application enables energy-efficient operation of O-RAN-based 5G campus networks. Using AI and machine learning, it analyzes and controls network operation in real time, making targeted adjustments to network elements or switching them off without compromising the quality of service.
