Verification and Validation Framework for Autonomous Rendezvous Systems in Terminal Phase

T HIS Note reports results from a study carried out for the ESA with the objective of improving the safety of future autonomous rendezvous guidance, navigation, and control (GNC) systems during the terminal rendezvous mission phase, recognized as a key capability for Mars sample return (MSR). The robustness of the critical terminal phase of the mission must be rigorously investigated under different conditions [1,2]. Availability of reliable verification and validation (VV) techniques, which can estimate the worst-case behavior of the system to provide guarantees of correct functionality with a desired safety level under different mission scenarios with a large number of uncertain conditions [3], is key to the success of the mission. In the case of collision avoidance, for example, the distance between chaser and target must be proven to always be greater than a specified minimum value even under worst-case conditions during the approach phase [1,4]. The most widely used VV technique in the space industry is still Monte Carlo (MC) simulation campaigns, which randomly explore the uncertain parameter space using highperformance simulators. Although easy to implement, key drawbacks are computational complexity and lack of guarantee to assess the trueworst-case (rare event) behavior of the system; see the results in [5] for an example of this phenomenon in the context of reusable launch vehicles. The contribution of the Note is an integrated approach combining the analytical μ analysis and the simulation-and-optimization-based method [5], which includes the global optimization algorithms such as Differential Evolution (DE) and Dividing Rectangles (DIRECT), as well as local optimization algorithm Nealder–Mead simplex [6]. The worst-case behavior of an autonomous rendezvous system, based on the industry standard high-integrity autonomous rendezvous and docking (HARVD) andGNC system [7] during the terminal rendezvous phase of a realistic MSR mission scenario, obtained by the proposed approach reveals the significant potential of the methodology when compared with traditional Monte Carlo simulations, in terms of both reliability and efficiency.