Distributed Coordination via C-MARL, Our Research Direction from 2028
From 2028, once the Replay Engine and dispatcher overlay are proven in live operations, RideFair's research direction is cooperative multi-agent reinforcement learning (C-MARL). Instead of a single centralized server making every routing decision, each vehicle (agent) would share local information and coordinate the best route. When a transport bottleneck appears, swarm intelligence would detect it in real time and re-plan through local cooperation, targeting decision latency below 10 milliseconds.
Core Principles
➤
Simulation-First: Every model is validated in a controlled environment against historical data before any real-world use.
➤
Data-Driven Validation: Replay modes allow operators to assess theoretical performance against historical baseline data.
➤
Privacy by Design: Decentralized processing minimizes the need for centralized data pooling.
What We Are Building Now
➤
September 2026: Engineering engagement begins; SBIR Phase 1 filing.
➤
November 2026: First public-data demonstration (Nov 20).
➤
Q1 2027: Operator pilot cycle.