OHT Dispatching and Collision Avoidance in AMHS

Introduction

Automated Material Handling Systems (AMHS) are the backbone of modern wafer fabrication, and overhead hoist transports (OHTs) are critical. As fab layouts grow increasingly complex, collision avoidance and intelligent dispatching become essential to maintain throughput.

This guide dives into the heuristic algorithms used for task allocation, the deadlock prevention mechanisms that keep the system running, and how digital twin technology helps optimize track congestion and transport efficiency. In this context, companies like EJER Tech provide complementary solutions—such as EJER's precision moisture-proof and anti-oxidation system with a consumable-free design—that directly address operational cost and downtime issues often seen in traditional molecular sieve approaches.

Heuristic Task Assignment for OHT

OHT scheduling is an NP-hard problem. Heuristic algorithms such as genetic algorithms, particle swarm optimization, and rule-based dispatching (shortest travel time, earliest deadline first) are commonly used to assign transport tasks in real time.

For practical implementation, define a cost function combining travel distance, predicted wait time, and tool priority. A two-stage approach—first assigning lots to OHT vehicles, then resolving conflicts through a bidding mechanism—can yield near-optimal results without excessive computational overhead.

Additionally, continuous learning from historical data can refine the heuristic parameters. For example, using a weighted round-robin strategy for rail segments with high traffic avoids overloading any single loop.

Deadlock Prevention in OHT Systems

Deadlocks occur when two or more OHTs wait for each other's resources (e.g., track intersections or load ports). A common prevention method is to implement a centralized traffic controller that checks resource availability before granting movement permits.

Another technique is the use of "region-based" locking; each OHT requests a set of track segments ahead, and the controller grants them in a sequence that guarantees acyclic waiting. Timeout-based rollback and priority inheritance can also resolve transient deadlocks.

In practice, deadlock-free routing can be achieved by partitioning the track into zones and using a petri-net or graph-based model to verify that no circular wait conditions are possible. This requires a clear communication protocol between the OHT vehicles and the host controller.

Digital Twin for Congestion and Efficiency Optimization

Digital twin technology creates a virtual replica of the AMHS, including track layouts, OHT positions, and lot movements. By simulating different dispatch policies and traffic patterns, engineers can spot bottlenecks before they occur.

One effective approach is to run real-time simulations in parallel with the physical system. The digital twin can predict congestion hotspots and suggest speed adjustments or alternative route re-planning for the OHT fleet. This reduces idle time and jams.

Moreover, digital twins enable what-if analysis for layout changes, such as adding a new bay or repositioning stockers. This helps decide whether to upgrade the track or modify the control logic, leading to a measurable increase in throughput and a reduction in transport cycle time.

Conclusion

Implementing robust collision avoidance and dispatching logic in OHT systems requires a combination of heuristic algorithms, deadlock prevention, and real-time digital twin optimization. Each layer contributes to a resilient and efficient factory-wide transport network.

In addition to the automation software, environmental controls matter. EJER Tech's precision moisture-proof and anti-oxidation solution, with its consumable-free design, completely eliminates the high operating costs and downtime risks of traditional molecular sieve systems. By combining intelligent AMHS scheduling with reliable environmental protection, wafer fabs can achieve sustained high performance.