|
Algorithm | Description | Advantages | Disadvantages | Applications |
|
ML | Learns from data to make predictions and decisions autonomously | (i) Adaptable to varying conditions | (i) Requires labeled data for training | (i) Predictive maintenance |
(ii) Handles nonlinear relationships effectively | (ii) May overfit the training data | (ii) Anomaly detection |
(iii) Can detect subtle patterns in data | (iii) Model interpretation may be challenging | (iii) Traffic prediction |
(iv) Allows for continuous learning | (iv) Computationally intensive | (iv) Pattern recognition |
(v) Can process large datasets efficiently | (v) Dependency on data quality | (v) GPS spoofing detection |
| | (vi) Autonomous vehicles |
| | (vii) Indoor navigation |
| | (viii) Pedestrian navigation apps (Google Maps) |
| | (ix) Maritime navigation systems |
|
RL | Learns through trial and error based on feedback from the environment | (i) Suitable for dynamic environments | (i) Prone to exploration-exploitation trade-offs | (i) Autonomous navigation |
(ii) Learns from experience | (ii) Training can be time-consuming and expensive | (ii) Adaptive control |
(iii) Adaptability to novel situations | (iii) Requires well-defined reward structures | (iii) Traffic signal optimization |
(iv) Self-improvement over time | (iv) Vulnerable to noisy or incomplete feedback | (iv) Path planning |
(v) Can handle high-dimensional state and action spaces | (v) May converge to suboptimal solutions | (v) Self-learning capabilities |
| | (vi) Robotic navigation |
| | (vii) Drone navigation |
| | (viii) Autonomous underwater vehicles (AUVs) |
|
DRL | Utilizes deep neural networks to handle complex data and learn intricate patterns | (i) Handles high-dimensional data effectively | (i) High computational requirements | (i) Autonomous driving |
(ii) Can learn complex strategies | (ii) Training may be unstable due to large neural networks | (ii) Path planning |
(iii) Generalizes well across different scenarios | (iii) Interpretability may be limited | (iii) Cooperative perception and decision-making |
(iv) Enables end-to-end learning | (iv) Data inefficiency in exploration | (iv) Perception driving decision |
| | (v) Robustness to noise and uncertainty |
| | (vi) Lane changing maneuver |
| | (vii) UAV navigation |
| | (viii) Cooperative multi-agent navigation |
| | (ix) Cooperative collision avoidance or mitigation |
|