The rapid proliferation of stochastic renewable generation and electric vehicle (EV) charging loads necessitates advanced operational frameworks that reconcile probabilistic grid constraints with adaptive control strategies. This paper presents some novel methodologies of integrating probabilistic constrained load flow (PCLF) analysis and copulas with deep reinforcement learning (DRL) and Variational Autoencoders (VAEs) to optimize EV charging schedules while maintaining grid reliability. By leveraging PCLF’s ability to quantify voltage and thermal limit violation probabilities under uncertainty and DRL’s capacity for sequential decision-making in high-dimensional state spaces or VAEs modified to use copulas in the latent space, the proposed framework enables dynamic, risk-aware EV charging coordination. Case studies on modified test systems demonstrate 12–18% improvements in voltage constraint satisfaction probabilities compared to deterministic approaches while reducing peak demand by 23%.

The increased adoption of Electric Vehicles (EV) necessitates smart and robust EV Supplying Equipment (EVSE) that can reliably charge EVs. Reliable EV charging means offering the right amount of Electric Power given the Power that the Energy grid can give at each time and also based on the Vehicles and Charging Point’s characteristics. A key problem that is observed, especially in Mediterranean countries with hot summers, is the high temperature values that are observed during charging. Very often, the maximum temperature is exceeded causing faults in the EVSE hardware, power losses, and interruption of the charging sessions. This paper proposes a smart framework to monitor the temperature of the EV and the EVSE, from a variety of inputs that are then fused together (sensor fusion), and pipelined to a Federated Learning algorithm to avoid hardware faults and suggest the optimal power so that the charging session continues reliable within safe operating temperature limits. The sensor fusion algorithm runs locally at each charging point, the results are then pipelined to a local Machine Learning algorithm and then the model parameters are sent via 5G to a cloud central FL model.

Virtual reality simulation of robotic-assisted min-imal invasive procedures reveals interesting issues related to the perception, control and manipulation of laparoscopic tools with emphasis given to the pivot trajectories and the Remote-Center-of-Motion (RCM) constrained motion planning. In this paper, the Gazebo simulator under Robot Operating System (ROS) allows the inclusion of hardware-in-the-loop for Minimally Invasive Surgery (MIS) procedures. The RCM constraint is addressed through the transformation of the surgical task space into the robot's taskspace, while addressing the robot's manipulability. Emphasis is given in calculating various geometric paths to be followed by the robot during surgery. Simulations were conducted using the ROS framework and the MoveIt kinematic planner using the RRTConnect path planning algorithm to evaluate the efficacy proposed scheme.