MUMBAI, India, Oct. 5 -- Intellectual Property India has published a patent application (202641115372 A) filed by Sri Eshwar College Of Engineering on September 24, 2026, for An Interpretable Multi-Agent Graph Planning Via Safety Interval Boundaries In Autonomous Driving.
Inventors include G. Dency Flora; R. Megala; G. G. Sreeja; and R. Arun.
The application for the patent was published on October 02, 2026, under issue no. 40/2026.
Abstract: Autonomous driving in complex and dynamically changing environments requires planning systems that can simultaneously reason about multiple interacting road agents while maintaining safety, efficiency, and interpretability. This work proposes an Interpretable Multi-Agent Graph Planning framework via Safety Interval Boundaries for autonomous driving. The proposed methodology represents the driving environment as a dynamic interaction graph, where the ego vehicle, surrounding vehicles, pedestrians, and relevant road elements are modeled as nodes and their spatial, temporal, and behavioral relationships are represented as edges. Multi-agent motion prediction is first performed to estimate the future trajectories and possible behaviors of surrounding agents. These predictions are then incorporated into the interaction graph to identify potential conflicts and interaction relationships. A major feature of the proposed approach is its intrinsic interpretability. Instead of treating safety as an implicit constraint within a black-box planner, the framework associates each planning decision with explicit safety-interval boundaries and interacting agents. Consequently, the system can explain why a candidate trajectory is feasible or infeasible, identify the active constraints responsible for a decision, and provide a concise rationale for the selected maneuver. The resulting closed-loop architecture continuously updates perception, prediction, graph relationships, safety intervals, and planning decisions as the environment evolves. The methodology therefore aims to provide a unified framework for safe, efficient, multi-agent-aware, and explainable trajectory planning in autonomous driving.
Disclaimer: Curated by HT Syndication.
