Remotely controlled vehicles (RCV) are mobile platforms that are moving and operating in distance to a human operator who is not in the vehicle. RCVs have a large range of applications with a huge market and cover remotely controlled robots (ground-based, underwater and unmanned aerial vehicles), ships, cars, spacecraft etc. In this project, the focus is on the rapidly growing segment of small RCV and small remotely piloted aerial systems (RPAS) in particular.Safety is a basic requirement for any application of RCVs and especially comprises a collision-free motion and action of the vehicle in the mission area. However, because of the limited situational awareness or the limitations of the data link, it is a critical task for the remote operator to avoid collisions with stationary and especially moving obstacles, especially in very complex scenarios such as urban environments.This leads to the idea to develop the required capabilities for automatically sensing the RCV’s surrounding and, if necessary, performing a suitable manoeuvre for collision avoidance as an emergency assistance system for the remote operator. Such systems are called sense-and-avoid systems and are considered as essential for the use of small RCVs, but also for larger RCVs and autonomous systems in general (such as autonomous cars for instance). In this project the focus is RPAS as one of the most challenging types of small RCVs, but all derived algorithms and solutions are in principle also applicable to the full class of small RCVs. One major objective of this project is the design of a robust but very flexible and extendable perception (software) architecture capable to detect the specific environment situation and autonomously select the most suitable sensor/algorithm configuration for sense-and-avoid solutions for small RPAS. Based on the identification and predicted motion of obstacles, an emergency evasive flight maneuver must be generated and effectuated onboard in real-time without exceeding the dynamic limits of the UAV and in combination with the commands of the remote operator. Therefore, another objective of this project is the investigation of a very flexible, robust but also real-time capable algorithmic approach for the generation of safety evasive emergency manoeuvres for collision avoidance of small RPAS even during remote control.