Much of the benefits of deploying unmanned aerial vehicles can be derived from autonomous missions. For such missions, however, sense-and-avoid capability (i.e., the ability to detect potential collisions and avoid them) is a critical requirement. Collision avoidance can be broadly classified into global and local path-planning algorithms, both of which need to be addressed in a successful mission. Whereas global path planning (which is mainly done offline) broadly lays out a path that reaches the goal point, local collision-avoidance algorithms, which are usually fast, reactive, and carried out online, ensure safety of the vehicle from unexpected and unforeseen obstacles/collisions. Even though many techniques for both global and local collision avoidance have been proposed in the recent literature, there is a great interest around the globe to solve this important problem comprehensively and efficiently and such techniques are still evolving. This paper presents a brief overview of a few promising and evolving ideas on collision avoidance for unmanned aerial vehicles, with a preferential bias toward local collision avoidance.

No abstract

No abstract

Inspired by the philosophy of aiming point guidance in the missile guidance literature, two new nonlinear guidance laws, named as 'nonlinear geometric guidance' and 'differential geometric guidance', are proposed in this paper for reactive obstacle avoidance of UAVs. A collision cone approach is used to predict any possible collision with the obstacle and, if necessary, to compute an alternate aiming direction for the vehicle. The proposed guidance strategies attempt to quickly align the velocity vector of the vehicle along the aiming point within a part of the available time-to-go, which ensures quick reaction and hence safety of the vehicle. Mathematical correlations between the guidance laws have also been established, which shows that the nonlinear differential geometric guidance and nonlinear geometric guidance are exactly correlated to each other with appropriate gain selections, while the linear aiming point guidance can only be approximately equivalent to the nonlinear differential geometric guidance. Using an alternate similar approach in place of the collision cone philosophy, these algorithms have also been extended for collision avoidance with moving obstacles in both non-cooperative as well as cooperative environments. The algorithms developed have been validated from a number of simulation studies in both two-dimensional as well as three-dimensional scenarios. Results are also included with an autopilot lag compensation logic to make the guidance laws more realistic. NomenclatureX v vehicle position, m V vehicle velocity, m/s X ap aiming point, m t go time-to-go, s X r relative distance between UAV and obstacle, m β safety circle around the obstacle in the plane containing X r and V r 1 , r 2 tangents from UAV CG to the safety circle β d radius of the safety circle of obstacle u i , i = 1, 2 unit vectors joining the center of β to r i t 0 initial time, s X d destination point, m (u, v, w) velocity of vehicle along body (x, y, z) axis respectively, m/s (v * , w * ) desired velocity along body (y, z) axis respectively, m/s (θ y , θ z ) aiming angle in (XY, XZ) planes respectively, rad ((X ap ) XY , (X ap ) XZ ) projection of the aiming point on to (XY, XZ) planes respectively, m c 1 , c 2 projection constants (k v , k w ) DGG gains in (XY, XZ) planes, sLGG and NGG gains in (XY, XZ) planes, m/s 2 e v error in y-velocity, m/s (τ v , τ w ) time constant for (y, z) velocity settling respectively, s (T s v , T s w ) settling time for (y, z) velocity settling respectively, s (α, β) settling ratio for (y, z) velocity settling respectivelyrelative distance between UAV1 and UAV2, m R 0 initial separation between UAV1 and UAV2, m V i velocity of UAV i , i = 1, 2, m/s t c time of closest approach, s r m miss distance between UAV1 and UAV2, m r saf e predefined safety distance between UAV1 and UAV2, m V des i desired total velocity of UAV i , i = 1, 2, m/s r vs i vector sharing displacement for UAV i , i = 1, 2, m r res distance to resolve conflict, m a * y

No abstract