We present the first demonstration of a fully spiking actor-critic neural network policy, trained via Proximal Policy Optimization (PPO), for continuous control of an agile high-speed quadcopter in a gate-based navigation task. The spiking neural network (SNN) controller employs Leaky Integrate-and-Fire neurons with surrogate gradient training and spike-rate decoding over multiple integration cycles, and it is benchmarked against a comparable artificial neural network (ANN) controller in both simulation and real-world flight tests. Results show that despite being trained to the same reward level, the SNN achieves superior performance in simulation, achieving higher episode rewards, greater robustness and reduced crash rate. Additionally, in 12-second real-world trials, the SNN outperforms the ANN, attaining a higher average reward (70.63 vs 59.77), greater mean velocity (7.94 vs 6.99 m/s), and more gates cleared (46.33 vs 40.67). An analysis of the spike integration cycle count reveals a clear trade-off: lower cycle counts (fewer integration steps per control update) reduce control output resolution and hinder learning, whereas higher cycle counts improve smoothness but increase inference latency. Moderate cycle counts (5 or 8) provide the best balance, yielding high rewards, smoother outputs, and low execution time overhead. These findings represent a key step forward for neuromorphic control in embedded autonomous systems, demonstrating that SNN-based policies can outperform conventional ANN controllers in high-speed, agile robotic tasks.

The final report from Delft University of Technology outlines the TurbEye project, focusing on the development of an autonomous drone designed for the inspection and maintenance of offshore wind turbines. The project was conducted by a team of students with the objective of creating a drone capable of performing maintenance checks through various Non-Destructive Testing (NDT) methods.

The project targeted the offshore wind energy market, which is growing rapidly compared to the more saturated onshore market. Offshore wind farms are increasingly significant due to their larger turbine sizes and greater capacity for energy production. These larger turbines require more efficient and automated inspection methods, which is where the TurbEye drone comes into play.

The TurbEye drone is designed to be fully autonomous, using a hydrogen propulsion system to achieve a longer range and endurance. The drone can cover a range of 280 kilometers and fly for up to 3.5 hours on a single hydrogen tank, inspecting up to six wind turbines per trip. This extended range and autonomy are crucial for reducing the high costs associated with manual inspections and human intervention, which currently dominate the industry.

The market analysis revealed that the offshore wind energy sector is expanding, with significant investments and a projected increase in capacity. The analysis also highlighted the need for innovative inspection solutions to manage the growing number and size of turbines. The TurbEye team identified a market gap for hydrogen-propelled autonomous drones, which offer advantages in terms of efficiency and cost-effectiveness over current manual and semi-automated inspection methods.

The drone’s system comprises several subsystems: structures, propulsion, control, and inspection. The propulsion system includes a hydrogen tank, fuel cell, backup battery, and propellers, selected for their safety and efficiency. The structural design uses lightweight and strong materials, ensuring the drone's robustness. The control system integrates advanced sensors and algorithms for autonomous navigation and inspection, while the inspection system employs a combination of visual cameras, passive thermography, and geometry inspection tools to detect surface and sub-surface damage.

Operationally, the drone follows a five-stage process during missions, from pre-flight checks and sensor calibration to data retrieval and refueling post-inspection. This systematic approach ensures the drone operates efficiently and safely. The planning of inspection routes is optimized using the Vehicle Routing Problem (VRP) algorithm, tested on the Hornsea 2 wind farm, the world’s largest. This optimization minimizes the number of trips and fuel consumption, enhancing operational efficiency.

AI algorithms are implemented to analyze the inspection data, detecting damage and differentiating between dirt and actual damage. These algorithms were trained on publicly available datasets but require further refinement due to limited training data.

The financial analysis estimates the production cost of a single drone at around 60,000 euros, driven mainly by the cost of the fuel cell, 3D scanner, and cameras. The financial projections suggest a total revenue of 9.77 million euros over five years, with a return on investment of 30.8%.

The project also emphasizes sustainability, aiming to minimize environmental impact through the use of green hydrogen and recycled materials, and aligning with the United Nations' Engineering for Sustainable Development framework.

Future steps involve further improvements in control systems, AI model reliability, detailed CAD designs, and dynamic simulations of the propulsion system. Extensive testing of subsystems and the complete system will follow before moving into production and operational phases. The ultimate goal is to attract customers and perform efficient and cost-effective wind turbine inspections, thereby reducing downtime and maintenance costs for offshore wind farms.