Deep Reinforcement Learning (DRL) shows great potential for flight control, due to its adaptability, fault-tolerance, and as it does not require an accurate system model. However, these techniques, like many machine learning applications, are considered black-box as their inner workings are hidden. This paper aims to break open the black box of RL for adaptive flight control by applying Shapley Additive Explanations (SHAP). The generated explanations are aimed at control experts, but can be useful for anyone interested in RL for adaptive flight control. This research proposes a novel Constant Weight Segment Detection (CWSD) algorithm, facilitating the use of eXplainable Artificial Intelligence techniques to adaptive RL. The algorithm and its usefulness are tested on an Adaptive Critic Design controlling a high-fidelity model of a Cessna Citation aircraft. It is demonstrated that SHAP in combination with CWSD provides detailed and useful insights into the relation between input and output of the RL algorithm. Using SHAP, linear relations between input and output are discovered, simplifying the understanding of the learned strategy.

The first launch of the Ariane 6 launch vehicle is planned for 2020, however in its current design no significant part of the launcher will be reusable. A current trend in the global space market is decreasing the costs of spacecraft launches through recovery, retrieval and refurbishment of parts of launchers. As a first step towards this market demand, it is to be investigated whether it is cost-effective to recover, refurbish and reuse the key components of the first stage of the Ariane 6, which are contained in the Vulcain Aft Bay (VuAB). This is where the engine, fuel lines, thrust frame and electronics are attached. The team has the task to develop a cost effective way of reusing the Vulcain Aft Bay. In the preceding report, multiple concepts were analysed and one concept was selected to complete the conceptual design. This concept consists of an Inflatable Heat Shield for re-entry, a Parafoil to control the flight at lower altitudes and a Mid-Air retrieval using a helicopter to perform a soft landing. A functional analysis was performed to define concept specific functions. This was done by means of a Functional Flow Diagram and Functional Breakdown Structure. In order to fulfil these functions, simulations were created of the most critical moments of the mission. One set of simulations analyse the trajectory of the system throughout the mission, predicting the location of landing. The other set of simulations is used to predict the critical load cases of the system. The aforementioned simulations were used to design the individual components of the system. By integrating these simulations and managing the iteration process, the overall system characteristics and configuration were established. The system was then analysed for sustainability, reliability, risk, maintainability and safety. The requirement compliance of the system was then updated, detailing which requirements have achieved full compliance, marginal compliance, no compliance and which have not been sufficiently investigated. Proposals to make all requirements fully compliant are included in a feasibility analysis. These include different design approaches for the team and design changes to the VuAB to accommodate the recovery system. From there strategies on future verification and validation activities was set forth together with operational, refurbishment and production plans. These plans and the design of the system were used to create an updated business model with return on investment figures, which predict a significant cost reduction on a per launch basis within five years. Finally a set of future recommendations and plan is proposed for the continuation of the project.