Table 1.
A summary of surveyed local motion planning methods for UAVs.
| Refs. | Control Structure | Local Motion Planning | Model | Dynamic Environment |
|---|---|---|---|---|
| [55] | I/II | sampling-based path planning | 2D Kinematics (nonholonomic) | ✓ |
| [56] | I/II | sampling-based path planning | 3D Single-rotor Dynamics | |
| [57] | I/II | sampling-based path planning | 3D Kinematics (nonholonomic) | ✓ |
| [58] | I/II | sampling-based path planning | 3D Kinematics (holonomic) | |
| [60] | I/II | graph-based path planning | 3D Kinematics (holonomic) | ✓ |
| [61,63] | I/II | optimization-based path planning | 3D Kinematics | |
| [62] | I/II | optimization-based path planning | 3D Quadrotor Dynamics | |
| [59,68,70,72,73,88] | II/III | optimization-based trajectory generation using QP with corridor-like constraints | 3D Quadrotor Dynamics | |
| [78,82,93] | III | optimization-based trajectory planning using QP | 3D Dynamics (acceleration/jerk input) | |
| [69,85] | III | optimization-based trajectory planning using unconstrained QP | 3D Quadrotor Dynamics | |
| [71,74,75] | III | optimization-based trajectory planning with obstacles constraints | 3D Quadrotor Dynamics | |
| [77,81,90,92,101] | III | motion primitives | 3D Quadrotor Dynamics | |
| [94] | III | motion primitives | 3D Kinematics (holonomic) | |
| [91] | III | motion primitives | 3D Kinematics (nonholonomic) | |
| [79,80] | III | perception-aware trajectory planning | 3D Dynamics (jerk input) | ✓ |
| [101,102,103,104,105] | III | perception-aware trajectory planning | 3D Quadrotor Dynamics | |
| [96,100] | III/IV | nonconvex optimization with obstacles constraints using NMPC | 3D Quadrotor Dynamics | ✓ |
| [83,84,86] | III/IV | mapless vision-based trajectory planning using depth images | 3D Dynamics (jerk input) | |
| [115,116,117,118,119,120] | IV | Geometric-based (collision cones) reactive control | 3D Kinematics | ✓ |
| [125,126] | IV | reactive control based on Velocity Obstacle (VO) | 3D Kinematics | ✓ |
| [127,128,129] | IV | reactive control based on artificial potential field | 3D Kinematics (nonholonomic)/Quadrotor Dynamics | ✓ |
| [121] | IV | nature-inspired reactive control | 3D Kinematics (nonholonomic) | |
| [134] | IV | vision-based reactive control | 2D Kinematics | |
| [131,132,133] | IV | real-time path deformation (reactive) | 3D Quadrotor Dynamics | ✓ |
| [135] | IV | vision-based reactive control based on NMPC | 3D Quadrotor Dynamics | ✓ |
| [136,137,138,139,140,141] | IV | deep reinforcement learning | 2D Kinematics | |
| [142] | IV | deep reinforcement learning | 2D Kinematics | ✓ |
| [143] | IV | deep reinforcement learning | 3D Kinematics | |
| [144,145,146,147,148] | IV | deep neural networks | 2D Kinematics | |
| [149,150] | IV | deep neural networks | 3D Kinematics | |
| [151] | IV | deep neural networks | 3D Kinematics | ✓ |