European Commission logo
español español
CORDIS - Resultados de investigaciones de la UE
CORDIS

GPS-free, beyond the visual line of sight navigation for logistics drones in urban environments

Periodic Reporting for period 1 - AUTOFLY (GPS-free, beyond the visual line of sight navigation for logistics drones in urban environments)

Período documentado: 2022-04-01 hasta 2022-11-30

Widespread drone adoption in Europe, under the right conditions and given the necessary technological developments, can help reduce Greenhouse Gas (GHG) emissions and improve the overall quality of life in European cities. To make commercial drone services feasible within urban areas, the EU has developed the concept of “U-Space”: “specific procedures designed to support safe, efficient, and secure access to airspace for large numbers of drones.” This is intended to enable the safe, large-scale deployment of drones at low altitudes within crowded urban environments but requires significant automation of functions. The current model requires human pilots to maintain visual contact with drones; pilots must still closely monitor flight operations and take over in the event of GPS loss, hardware malfunction, battery failure, bad weather, or ground obstacles. A human pilot also needs to ensure drop point areas are clear when performing delivery missions. These factors severely limit the scalability of drone operations and prevent their safe implementation in dense urban environments.

The rise of e-commerce has resulted in increased last-mile logistical complexity, as individual deliveries have increasingly become the norm. ~ 50% of total logistics costs (fuel, fleet operation costs, wages, etc.) are in the last mile. Last-mile deliveries using ground vehicles are subject to unpredictable factors, such as traffic jams, inclement weather conditions, and road closures, and carry a meaningful carbon footprint. Outside cities, same-day delivery options are limited. Drone deliveries will enable same-day service (within hours or minutes), even in rural areas and congested urban environments, with a lower carbon footprint. Drones can also improve the transport of sensitive goods such as medical goods and legal documents. (Ernst and Young, 2021; Forbes, 18 August 2021).Public utilities must expend significant manpower on inspecting infrastructure. While drones are currently used to some extent, they must now be operated within an operator's Visual Line Of Sight (BVLOS), thus significantly reducing their productivity and their ability to save costs.

Today, there is no affordable, lightweight solution that can enable drones to make autonomous decisions beyond the operator’s line of sight, to position themselves accurately in 3D space, even when GPS is unavailable or not or accurate enough, to avoid hazards by detecting and evaluating objects in their environment, and to precisely choose safe areas for landing and delivery, without risking people or property. There is not yet a solution on the market which can reliably provide (among other things) ‘detect and avoid’ capabilities, geolocation services and emergency management in all circumstances (including in high complexity, dynamic environments where GPS may not be reliable).

This project will validate, qualify, and commercialize Sightec’s safety solution, which represents an enabling technology helping to unlock the full potential of drones for Europe and the world. Using cutting-edge Computer Vision and Artificial Intelligence (AI) algorithms, which provides vision-based navigation, safe landing, obstacle avoidance and safe delivery, Sightec's software turns the drone's camera into a smart and affordable sensor - enabling the drone to “see and understand its surroundings” like a human pilot.
R&D activities within this project focus on finalization of algorithm development to enable full functionality of our minimum viable product (MVP) daytime navigation system under all operational conditions. It also allows easy integration of our system with commercially available autopilot systems, will develop an interface for remote troubleshooting, enable over-the-air software updates, develop a mapping server (with GUI) to allow automatic server distribution of mapping data to client drones, and perform other essential upgrades to enable scale-up and commercial deployment of a reliable Sightec product.

Validation and qualification activities conducted as pilots and demonstration (~4 such pilots) establishes our ability to conduct safe navigation, landing, and delivery in urban environments for logistics and autonomous inspection missions.

Within the period of this report (M1-M8), we have completed our positioning module for daylight (EO camera).

The objectives of this task were to:
• Improve registration of image vs orthophoto by using neural networks.
• Improve visual-inertial odometry, by better fusion with Inertial Measurement Units (IMU), and faster computation to achieve higher framerate, to sustain accurate positioning with minimal drift for longer periods of time.
• Improve standalone ability to loiter and descend vertically when absolute positioning is not possible, and to fuse this ability with IMU measurements to overcome cases where the prior information of the terrain is inaccurate.
• Conducting test flights to qualify system.
Successful achievement of the daylight positioning module has resulted in a system capable of providing accurate absolute positioning when possible and relative positioning when not. In the former case, the drone is able to complete its mission when GNSS signal is lost. In the latter case, the drone remains within its fly zone so it does not endanger its surroundings.

Further technical activities within the grant action will involve productization and infrastructure task which will enhance our ability to deploy to many customers and to increase product reliability once commercialized, and autopilot integration, which will transform the Sightec solution into a standalone sensor which can serve as a GNSS backup.