Table 10.

Multi robot and swarm intelligence algorithms (UUVs).

Problem	Resolution	Performance and Additional Explanation	Ref.
UUV
Multi-AUV cooperation method	End-to-end MARL (multiagent reinforcement learning)	Markov decision process for navigating. CT-DE (centralized training with distributed execution) for path planning Obtain data through equipped sonars, electronic compasses, and inertial sensors via the Markov decision process MADDPG (multiagent deep deterministic policy gradient) algorithm is used for the end-to-end AUV control algorithm	[169]
Multi-AUV cooperation method	Genetic algorithm	Possible cost-performance trade-off Simulate up to 3 AUVs Automatically recharge energy at stationary charging stations The trajectories and positions of the AUV and charger are generated after utilizing the genetic algorithm as a global optimization too	[170]
Multi-AUV cooperation method and obstacle avoidance	Bio-inspired neural network algorithm	Bio-inspired neural network algorithm is used for path planning Shorter length of the trajectories than that of the artificial potential field method A 3D grid-based active model expressed as a bio-inspired neural network algorithm Simulation is conducted with conditions such as the presence of obstacles and different densities of obstacles	[171]
Multi-AUV cooperation method and network architecture	Underwater cooperative navigation technique based on SDN	Adaptive optimization policy for C-AUVs and predefined fixed spiral elliptic trajectory from top to bottom for S-AUVs are sued. Centralized network management Good performance in terms of execution efficiency and system stability Easier to deploy and more efficient in planning the AUV’s cruising trajectory	[172]
Route planning	Hybrid path planning	Shorten algorithm execution time and elimination of nonexecutable paths Detect obstacles using multibeam forward-seeking sonar (FLS) and create outlines (polygons) of obstacles Hybrid path planning algorithm based on PSO and waypoint guidance	[173]
Route planning	SAC (soft actor–critic) algorithm	dynamic detection scheme is used for path planning C: SDN (Software-Defined Networking) controller underwater diffusion source route planning for Pollution Detection Leading the Paradigm of Multi-AUV Network Intelligent Transportation Systems (SDNA-ITS)	[174]