Creating the requirements for safe autonomous vehicles
Developing safer, greener, more secure, quiet and accepted Urban Air Mobility is challenging due to many factors, such as environment, regulations, and safety-critical technologies.
AURORA tackles these challenges by developing and integrating safety-critical technologies to support autonomous flight UAM in urban environments, focusing its demonstrations primarily on emergency-related applications. Jean-Emmanuel Haugeard (Thales) and Ignacio Querol Puchal (SEALA) are two of our team members responsible for bringing these technologies to life.
We sat down with the two engineers to discuss their experiences in creating AURORA’s building blocks over the past year, trying to understand their goals, obstacles and lessons learnt.
Creating the building blocks for autonomous navigation
The technologies being developed by Thales offer a state-of-the-art industry view on perception solutions for autonomous systems. With their expertise and resources, Thales is actively contributing to AURORA with the development of vision algorithms.
Using the vehicle’s camera, photos available in a geo-referenced images database and the rough localisation of the vehicle, Jean-Emmanuel and his team are using computer vision and machine learning approaches to develop tools concerning autonomous navigation. These solutions will be able to estimate the current location of the unmanned vehicle and to extract semantic information of the ground in order to determine possible landing places. These efforts address the problematics of absolute localisation using visual perceptions in GPS denied areas.
In parallel, SEALA is developing an algorithm that would plan the global flight path of UAM vehicles. Together with the University of Ghent, Ignacio has developed an algorithm that allows to find a continuous path or trajectory that will drive the aircraft from a start to an end location knowing the environment map.
The final logical building-block is to identify a safe landing location, because an aircraft cannot continuously fly. With the visual information of Thales, the SEALA team is able to calculate a route that integrates the dynamics of the vehicle and therefore recognise a safe site to land.
The urban context and its difficulties
For both engineers, the urban context of AURORA poses some obstacles for their technical developments. Not only is the urban context of a higher complexity, requiring more creativity and ingenuity in creating the algorithms, but it is also not so heavily documented in data, making it more difficult, for example, to find which sites would be appropriate for landing.
The lack of diversity in the available data was a clear challenge both Jean-Emmanuel and Ignacio had (and still have) to overcome. This is particularly difficult for the urban context, where there is less data available. Urban areas are by definition diverse: no city is alike, they have buildings, rooftops, streets of all shapes and sizes, they are often overpopulated and have intense traffic. This diversity can make it challenging and time-consuming to transfer data onto the algorithms and to find appropriate routes and landing sites.
Lessons learnt and looking forward
During this year, our engineers learnt several lessons, overcoming the barriers posed and adding daily to innovation in Urban Air Mobility. Their constant curiosity and attention to detail made it possible for our team to get so far.
Stay curious: don’t consider just your part, but the overall solution. For example, a software solution could be relevant during the simulation phases, but may show low performance in a real situation. The causes could be attributable to software/hardware issues (for example: synchronisation of sensors). We should consider software and hardware as a global system. Mastering both is a key factor in developing a viable, robust and safe solution
Jean-Emmanuel Haugeard (Thales)
Looking back, they would recommend any young researcher in the same field to stay curious, to consider the whole picture and how the different pieces connect and to never give up, even when the technology seems to be against you.
Never give up! The method can seem really complex, but these kinds of problems need some high level solutions. Do not be afraid and keep up on your work that this is good enough, there’s always room for improvement.
Ignacio Querol Puchal (SEALA)
In the near future, the SEALA team will start the local flight path planning activities. With this we will progress in the path calculation by adding moving obstacles that can represent a more dangerous hazard to the vehicle. The objective here is to dynamically update the global path plan with current localisation information.
As the second year of AURORA progresses, these building blocks will increase in complexity so that they can ultimately be integrated and tested both in a virtual reality setting and in the real world in our pilot locations.