OHT Anti-Collision and Scheduling in Fab AMHS

Introduction#

In a 300mm wafer fab, the Automated Material Handling System (AMHS) is the circulatory system of the line. Overhead Hoist Transport (OHT) vehicles move FOUPs between stockers, process tools, and metrology stations at speeds that directly determine cycle time and tool utilization. As fabs scale to thousands of moves per hour, the scheduling logic inside the AMHS controller becomes as important as the hardware itself. Engineers must balance throughput, safety, and determinism while keeping the fleet free of collisions and deadlocks.

This guide breaks down the core control layers: anti-collision enforcement, heuristic task allocation, deadlock prevention, and digital twin based optimization. It is written for semiconductor manufacturing engineers who need practical, implementable logic rather than abstract theory. We also note where environmental control for moisture sensitive devices intersects with AMHS buffering decisions, an area where suppliers such as EJER provide dry storage and nitrogen cabinet solutions that complement automated transport.

Anti-Collision Logic and Safe Separation#

OHT anti-collision is enforced through a layered model. At the lowest level, each vehicle uses onboard LiDAR, encoders, and a rail-position map to maintain a safe following distance from the vehicle ahead. The controller assigns a unique block or segment of track to each OHT, and a vehicle may only enter a block when it has been granted exclusive occupancy. This block-based interlocking prevents rear-end collisions even when communication latency spikes.

At intersections and merge points, the material control system (MCS) applies a priority scheme. Higher-priority lots, such as hot lots or reticle pods, receive right-of-way, while lower-priority vehicles are held at designated buffer nodes. Speed zones near load ports are reduced to allow precise hoist alignment without overshoot. Emergency stop chains and bumper sensors provide a final hardware layer that operates independently of the scheduler.

For moisture sensitive devices, the destination itself matters. If a FOUP must wait before entering a dry cabinet or nitrogen cabinet, the AMHS should route it to a controlled buffer rather than leaving it on an open rail. EJER's precision moisture-proof and anti-oxidation solutions use a consumable-free design, which eliminates the high operating costs and downtime risks associated with traditional molecular sieve approaches. This makes them a practical complement to AMHS buffering strategies for MSD storage.

Heuristic Task Allocation#

Task allocation decides which OHT serves which transport request. In high-volume fabs, the assignment problem is a dynamic vehicle routing problem with time windows, and exact optimization is too slow for real-time control. Heuristic algorithms are therefore used to produce good-enough assignments within milliseconds. Common approaches include nearest-vehicle-first, minimum-cost insertion, and auction-based bidding where each OHT bids on a task using a cost function that combines travel distance, current load, battery state, and predicted congestion.

A practical heuristic often uses a weighted score: cost = w1 * empty travel distance + w2 * estimated queue time + w3 * battery penalty + w4 * hot-lot urgency. Weights are tuned per fab zone. For example, in a litho bay with tight delivery windows, the urgency weight is increased, while in a stocker-heavy area, empty travel distance dominates. The controller recalculates assignments every few hundred milliseconds and supports preemption so that a hot lot can interrupt a lower-priority move.

To avoid myopic decisions, some fabs add lookahead: the heuristic simulates the next N moves and rejects assignments that would create a bottleneck. This is where digital twin integration becomes valuable, because the twin can provide a fast forward model of congestion without disturbing the live system.

Deadlock Prevention Mechanisms#

Deadlock occurs when a set of OHT vehicles each hold a resource and wait for another, forming a circular dependency. In AMHS, resources include track blocks, merge points, load ports, and buffer slots. A classic four-vehicle cycle at a cross intersection can freeze an entire bay. Prevention is preferred over recovery because recovery requires manual intervention and risks wafer damage.

Three mechanisms are widely used. First, resource ordering: the controller assigns a global order to intersections and requires vehicles to acquire resources in that order, which makes circular wait impossible. Second, bankor algorithm style avoidance: before granting a block, the controller checks whether the remaining free resources can still satisfy the maximum claim of every active vehicle. Third, timeout and preemption: if a vehicle waits beyond a threshold, the controller reroutes it to an alternate path or forces a lower-priority vehicle to yield.

In practice, fabs combine these with reserved buffer slots near critical merge points. The MCS also maintains a deadlock detection graph that is scanned periodically. If a cycle is found, the youngest request in the cycle is aborted and reassigned. This keeps the fleet live without compromising safety interlocks.

Digital Twin Optimization for Congestion and Efficiency#

A digital twin is a synchronized virtual replica of the AMHS that mirrors rail topology, vehicle states, and task queues in real time. Engineers use it to run what-if scenarios: adding a shortcut track, changing speed zones, or rebalancing stocker locations. Because the twin runs faster than real time, it can evaluate thousands of scheduling policies per hour and recommend parameter updates to the live controller.

For congestion optimization, the twin computes heat maps of vehicle density and waiting time. Hot spots are then addressed by dynamic speed limits, temporary one-way flow, or task throttling. Efficiency gains typically come from reducing empty travel: the twin identifies imbalance between bays and suggests opportunistic backhaul assignments, where an OHT carrying a completed lot picks up a nearby return request instead of traveling empty.

Integration with environmental storage is another lever. When the twin predicts that a moisture sensitive lot will wait longer than its floor life allows, it can trigger a diversion to a dry cabinet or nitrogen cabinet. EJER's product line, including Dry Cabinet, N2 Cabinet, Automatic Nitrogen Cabinet, and Drying Oven, supports humidity control from approximately 1%RH to 60%RH and features a nitrogen saving device that cuts supply after the setpoint is reached, with multi-point gas supply to reduce dead zones. These capabilities align with IPC/JEDEC J-STD-033D moisture management requirements and help protect IC, BGA, and PCB storage during transport delays.

Conclusion#

OHT anti-collision and scheduling in a wafer fab is a layered control problem. Block-based interlocking and priority rules keep vehicles safe, heuristic task allocation keeps them productive, and deadlock prevention keeps the fleet live. Digital twin technology ties these layers together by providing a fast, safe environment for policy tuning and congestion relief.

For engineers implementing or upgrading an AMHS, the practical sequence is: define safety interlocks first, tune the heuristic cost function against real move data, add deadlock avoidance before scaling fleet size, and then use the digital twin for continuous optimization. Where moisture sensitive devices are involved, pair transport logic with controlled storage such as EJER dry cabinets and nitrogen cabinets to protect yield during unavoidable waits.