We consider the challenges discovered while transitioning a robot from teleoperated to autonomous. Our platform is a versatile ground robot with variable geometry, changing its width to allow for either increased stability or entry into smaller gaps. This additional articulation presents problems and opportunities with regard to the stability of the platform. We show that control over the track width allows us to maintain stability through the use of relative maximum allowable parameters. As the narrow track width causes the robot to become less stable, to keep the platform stable the system controlling it needs to respect these maximums.

In this paper, we perform an analysis of the fighter bomber duel where player one (the fighter) has a discrete resource, as in a single shot which he can fire at any point, and the other player (the bomber) has a continuous resource (machine gun), which she can distribute as a function of time. We look at this problem applied towards the safety of an autonomous system in a competitive environment. The specific problem we address is probability of failure, can we put a bound on the probability that an autonomous system completes its objective in a hostile environment?

Insects are well known to be adept at flying through cluttered natural environments. This ability to avoid collisions and control flight speed is an active field of interest for potential applications in unmanned aerial vehicles. Previous studies have shown that insects primarily rely on visual motion of the environment, known as optic flow, to perform flight manoeuvres such as course control, landing, terrain following, etc. Vision based flight behaviours of honeybees have been studied extensively in the past and have been explained in terms of either the optomotor response or the collision avoidance (centering) response. However, the strategies used for avoiding smaller objects in the frontal view field remain unclear. This study investigates the strategies being used by honeybees (Apis Milfera) to avoid such obstacles. This was done by performing behavioural experiments, where the bees were trained to fly in a tunnel and were then presented with cylindrical obstacles of six different sizes ranging from 25mm to 165mm. The flights were recorded using a GoPro camera and then digitised using Matlab. The digitised trajectories have then been analysed for cues such as retinal angle, relative retinal expansion velocity(RREV), optic flow, etc., to gain an insight into the visuo-motor strategies being implemented by honeybees to avoid these obstacles. Our findings, based on analysing major events during flight, such as the point of deceleration before the obstacle, the point of maximum curvature and the point where bees cross the obstacle, suggest a combination of RREV and optic flow based response to avoid these obstacles.