Fraunhofer Institute for Transportation and Infrastructure Systems IVI
Fraunhofer Institute for Transportation and Infrastructure Systems IVI

Autonomous Flying

Shaping the future of autonomous aviation

Advanced air mobility and autonomous flying will revolutionize the future of mobility and aviation. Shifting urban transportation into the air will open up new possibilities both for goods shipping and for sustainable personal transport.

While the autonomous flight of commercial aircraft at cruising altitudes on preplanned routes is already a sufficiently solved problem, completely new challenges arise in the context of advanced air mobility. In particular, flying in airspace close to the ground – i. e., taking off and landing as well as flying below 120 meters on non-predefined routes – raises new questions that show strong analogies to autonomous driving.

The research work at Fraunhofer IVI focuses in particular on the development of trustworthy, AI-based autonomy systems. The use of these systems, for example in the form of pilot assistance systems, promises immense application potential and will be crucial for the widespread use of new applications in aviation.

To meet these challenges, the “Autonomous Aviation Systems” group at the Fraunhofer Application Center “Connected Mobility and Infrastructure” has built up extensive expertise in the fields of artificial intelligence (AI), drone technology and control engineering. With our cross-disciplinary expertise in the hardware and software integration of drones as well as in sensor systems and embedded systems, we develop our own sensor platforms for collecting new types of data in the field of autonomous aviation.

Chart on autonomous flying
© Fraunhofer IVI
Overview of the framework of autonomous flying.

Expertise

Expand all Close all

  • Deep learning
    • Development, optimization and evaluation of high-performance deep learning models under real-time requirements for use in embedded systems
  • Computer vision
    • Object detection and semantic segmentation in 2D and 3D for ground and aerial applications
    • Occupancy grid prediction with the help of state-of-the-art AI architectures
    • Monocular depth estimation and 3D semantic evaluation of (infrared) camera, LiDAR and RADAR data through AI-based sensor fusion
  • Application of transfer learning and student-teacher approaches to working with small data sets with the aim of reducing model complexity
  • Holistic investigation of redundancy, explicability and reliability in compliance with aviation standards
  • Continuous optimization of the trustworthiness of AI systems through research in the fields of out-of-distribution detection, uncertainty estimation and local robustness

  • Reliability, safety and performance guarantee of cooperative multi-agent algorithms
  • Semantic path and trajectory planning with single-shot approaches
  • Runtime assurance algorithms based on control barrier functions for safeguarding procedures with a focus on avoiding collisions with obstacles and/or non-cooperative agents

  • Design and implementation of non-linear control methods such as gain-scheduling method, non-linear dynamic inversion and model-predictive control
  • Development of decentralized cooperative path planning algorithms based on model-predictive strategies
  • Development of cooperative path planning strategies based on token passing methods in the context of cooperative flight path planning

  • Methods for deterministic and stochastic systems including observers, Kalman filters and particle filters for use in navigation filters for self-localization and in tracking filters for the trajectory prediction for dynamic and extended objects
  • Investigation and improvement of methods for decentralized status estimation 

Spotlight

Safe Autonomous Landing in Unknown Terrain: A Vision-Guided and GNSS-Denied Solution.

(Markus Gross, Felix Soest, Philip Hausmann and Henri Meeß, Fraunhofer IVI).

Application areas

Autonomous (emergency) landing of drones

  • Development of functions for autonomous (emergency) landing in unknown terrain without the reliance on external communication, GNSS, and pre-planned landing sites
    • AI-based semantic environment perception
    • Integration of mechatronics and software into existing drones

AI-based flight and maneuver planning

  • Development and implementation of AI-supported algorithms for trustworthy and reliable localization, control, trajectory planning and environment perception
  • Use of methods from reinforcement learning and computer vision in addition to our own data sets to develop high-performance systems

© Marc Müller

Training and transfer

  • Customer-specific workshops on AI-based functions and technology including idea generation for concrete implementation 
  • Interactive and modular concept for SMEs and Startups from  mechatronics, prototyping, aviation, transportation, etc.

Selected publicly funded projects

Building on the Fraunhofer ALBACOPTER® flagship project, a comprehensive collection of projects in the section of highly automated flying was established at Fraunhofer IVI. These projects include topics such as AI-based environment perception, sensor systems for autonomous drones, path planning and further key technology areas. The sub-systems developed within the scope of these projects, such as semantic environment perception, semantic path planning and autonomous (emergency) landing, make major contributions to the autonomy of drones. We are working hard on coordinating several autonomous platforms and are developing distributed coordination algorithms and runtime assurance algorithms for decentralized trajectory planning.

 

Experimental Vertical Take-Off and Landing Glider

ALBACOPTER®

With their mission statement "Leading the world towards safe autonomous flying", the Fraunhofer IVI's Application Center has started their initiative for highly automated aviation within the scope of the Fraunhofer ALBACOPTER® flagship project. Its main goal is to develop a modular and universal autopilote suite for various scenarios in lower airspace.

More information

UAM – Reliable environment perception

VERUM

In the aviation section, the trustworthiness, reliability and certifiability of AI systems, especially for environment perception, represent a previously unsolved problem. The VERUM project aims at creating a reference architecture for reliable environment detection in unmanned air mobility (UAM). In this context, the Fraunhofer Application Center "Connected Mobility and Infrastructure" is developing AI algorithms for evaluating (IR) camera, LiDAR and Radar data for environment perception and is compiling a reference data set for highly automated flying.

© VanderWolf Images - stock.adobe.com

Energy-efficient flight control

ENGEL

In air rescue missions, the mission locations often lie in unknown terrain, where landing is not without risks for helicopters. 

The aim of the ENGEL project is to better detect and understand helicopters' environments during their cruising, landing approach and landing stages with the help of suitable sensor systems and artificial intelligence, and to ultimately enable safer and faster landing.

Non-cooperative aircraft collision prevention in low-altitude scenarios

NiKoLaS

Safe take-off and landing phases are integral for future partially and fully automated aircraft. For this purpose, centrally distributed information about other air traffic participants is often processed and exchanged during the flight phase. On and close to the ground, however, there are many passive objects and obstacles not included in the shared information. In the NiKoLaS project, solutions are created for detecting potential collision objects in the environment with the aim of enabling partially and fully automated aircraft navigation in low altitudes.

 

Advanced air space mapping

ADAM

Whether logistics drones, air taxis or rescue and surveillance systems – the requirements for flight safety in the urban air mobility (UAM) section are particularly high, regardless of the respective application.

The ADAM ("Advanced Air Space Mapping") project plans to improve environment perseption in lower airspace with the help of new mapping methods to enhance the safety of autonomous flying in urban settings.

ADAM at LinkedIn

MEDIC

Having medication available in hospitals and being able to ship lab samples and blood bags quickly is a logistics problem that could be solved with the help of drones. 

The goal of the MEDIC project is to develop an operating and flight guidance system for a single-rotor drone that enables intelligent flight control and autonomous landing in both drone hubs and unknown terrain.

IDEALS

The routes of future transportation networks will run through logistics hubs equipped with vertiports in which the air freight is loaded into autonomous drones. The vertiports' limited capacity represents a bottleneck in terms of high drone density in their airspace. As a preliminary study, the IDEALS investigated cooperative coordination in intelligent networks of multiple, heterogeneous and autonomous transport drones for take-off and landing procedures at the vertiports of logistics centers.

Autonomous and safe urban air mobility

ASAAM

The ASAAM project is an international cooperation between NTU in Singapore and the Fraunhofer Institutes IVI, IEM and IOSB. It intends to improve the scalability and safety of advenced air mobility (AAM). This includes the development of automated take-off and landing maneuvers that comply with the expected safety requirements of the EASA. The work focuses on redundant end-to-end raw data fusion and progressive algorithms for safe trajectory planning and robust flight control. 

News

© Kadmy - stock.adobe.com

Research flights in the Ingolstadt city area

Further information, events and contact persons will be posted here.

Note on data protection

As soon as the first test flights have been scheduled, you can read our data protection concept here.

 

© AININ

28.9.2024

Young talents in urban air mobility at the 4th Research Summer Camp

Together with the Artificial Intelligence Network Ingolstadt (AININ), Fraunhofer IVI's Application Center "Connected Mobility and Infrastructure" invited 20 students from all over Germany to the Ignaz Kögler Research Summer Camp in Ingolstadt. During the week, the participants immersed themselves in the subject of "Urban air mobility and autonomous flying". Their work focused on topics such as AI-based environment and situation perception, object detection, as well as trajectory prediction and intelligent trajectory planning.

read more
 

© City of Ingolstadt / Rössle

13.6.2024

UAM Initiative Ingolstadt network meeting

The network meeting offered fascinating insights into the latest research topics in Urban Air Mobility and featured practical outdoor demonstrations. The Fraunhofer flagship project ALBACOPTER®, including six sub-projects, was the highlight of the event.

read more

Publications

Jahr
Year		Titel/Autor:in
Title/Author		Publikationstyp
Publication Type
2024 Drones for automated parcel delivery: Use case identification and derivation of technical requirements
Zieher, Simon; Olcay, Ertug; Kefferpütz, Klaus; Salamat, Babak; Olzem, Sebastian; Elsbacher, Gerhard; Meeß, Henri		 Zeitschriftenaufsatz
Journal Article
2024 Collaborative Semantic Occupancy Prediction with Hybrid Feature Fusion in Connected Automated Vehicles
Song, Rui; Liang, Chenwei; Cao, Hu; Yan, Zhiran; Zimmer, Walter; Gross, Markus; Festag, Andreas; Knoll, Alois		 Konferenzbeitrag
Conference Paper
2024 Evolutionary algorithms for a simheuristic optimization of the product-service system design
Meeß, Henri; Herzog, Michael; Alp, Enes; Kuhlenkötter, Bernd		 Zeitschriftenaufsatz
Journal Article
2024 Dynamic Obstacle Avoidance for UAVs using MPC and GP-Based Motion Forecast
Olcay, Ertug; Meeß, Henri; Elger, Gordon		 Konferenzbeitrag
Conference Paper
2024 Wettbewerbsfaktor Luft- und Raumfahrt - zukunftsfähig mit Fraunhofer
Battenberg, Andreas; Ben Bekhti-Winkel, Nadya; Griesbach, Elke; Kothe, Simon Markus; Lauster, Michael; Lindow, Kai; Loosen, Thomas; Meeß, Henri; Pauly, Gerhard; Schäfer, Frank; Dietzsch, Josephine; Strohbach, Tim; Wansch, Rainer		 Paper
2024 First steps towards real-world traffic signal control optimisation by reinforcement learning
Meeß, Henri; Gerner, Jeremias; Hein, Daniel; Schmidtner, Stefanie; Elger, Gordon; Bogenberger, Klaus		 Zeitschriftenaufsatz
Journal Article
Diese Liste ist ein Auszug aus der Publikationsplattform Fraunhofer-Publica

This list has been generated from the publication platform Fraunhofer-Publica