Search for a command to run...
The authors of this research paper have developed an Artificial Intelligence predictive control technology to enhance the operational safety and performance of Unmanned Aerial Vehicles (UAV s). The central goal of the methodology is to allow a UAV to recognize when it will likely encounter a hazard before the hazard occurs so it may take a preventative measure. This methodology employs a predictive model utilizing both Deep Deterministic Policy Gradient (DDPG) and Model Predictive Control (MPC), to forecast future hazards, thus enabling the UAV to proactively take action to prevent future hazards from occurring. Simulation results showed that the UAV avoided many crash situations, and used less fuel while flying than traditional reactive systems of control. Additionally, the UAV's overall stability and reliability improved significantly with each additional flight, regardless of whether or not it encountered adverse weather and/or lost satellite GPS signals. Results obtained from the study support the concept that UAVs may safely operate using predictive control methodologies instead of only relying upon reactive control methodologies. Although the findings of the study indicate significant promise for predictive control methodologies, there remains opportunity for further improvements regarding performance. Nonetheless, the authors successfully demonstrated their methodology via successful testing. Ultimately, the authors aim to create UAVs that make decisions based on similar processes that humans use by enabling the UAV to develop plans of action prior to making contact with potential problems.