Twin Delayed Deep Deterministic Policy Gradient for Adaptive Autonomous Navigation with LiDAR and Camera Sensor Fusion in Diverse Environment

Intelligent robots are increasingly being deployed for autonomous navigation and complex task execution in diverse environments. However, the current applications of the twin delayed deep deterministic policy gradient (TD3) algorithm, utilizing sensor fusion data, fail to adequately address the complexities of indoor diverse environmental conditions and probability of identification (POI).In this research, we introduce an twin delayed deep deterministic policy gradient (TD3) algorithm that uses sensor fusion data from RGB and VLP-16 LiDAR cameras to perform well in challenging real-world scenarios. This improvement raises the algorithm’s level of spatial awareness and navigation accuracy by enabling it to collect extensive environmental data. The algorithm also makes use of sophisticated self-learning features, which allow it to precisely locate key POIs and adjust dynamically to changing environments. It is suitable for applications with limited resources because of these adaptive properties, which maximise both performance and computational efficenicy The enhanced model has been validated by extensive simulations and experimental assessments, demonstrating its great potential for enhancing the adaptability, dependability, and sustainability of autonomous systems in a variety of industries, from search and rescue to logistics and environmental monitoring.