Design Warehouse Robotics - System Design | WinJob

Problem Statement: Autonomous Warehouse Robotics

Phase understanding — Amazon/Ocado-class systems coordinate autonomous mobile robots (AMRs) that move inventory pods or totes from storage to pack stations while humans work nearby.

Concern	Decision
1	Accept pick tasks from WMS and assign idle robots with sufficient battery
2	Maintain a live occupancy map with reserved spacetime corridors per robot
3	Replan within 200–500 ms when obstacles, faults, or priority changes occur
4	Expose fleet health, throughput, and stuck-robot alerts to operators

Fleet charter (1)

Treat the control plane as a traffic manager for physical packets: each robot is a moving lock on grid cells. Interviewers probe whether you separate task assignment (which SKU, which robot) from motion planning (how it moves without collision).

Safety invariant: No commanded velocity without a valid reservation token for the next cell segment. Emergency estop propagates over a dedicated low-latency channel, not the task queue.

Throughput lens: Measure picks per robot-hour and aisle utilization; a path that is collision-free but causes convoys is still a failed design.

Edge cases: Robot loses localization in a mirror aisle — freeze reservations, request human assist, never guess coordinates. Duplicate MQTT pose frames must be idempotent on (robot_id, seq).

javaOne Dark Pro

1 public record PickTask(String taskId, String sku, GridCell source, GridCell station, int priority) {}

pythonOne Dark Pro

1 def cells_for_path(path: list[tuple[int, int]], t0_ms: int, speed_mps: float) -> list[tuple[tuple[int,int], int, int]]:
2     """Return (cell, enter_ms, exit_ms) reservations along a polyline."""
3     out = []
4     t = t0_ms
5     for a, b in zip(path, path[1:]):
6         dt = int(1000 * grid_distance(a, b) / speed_mps)
7         out.append((b, t, t + dt))
8         t += dt
9     return out

typescriptOne Dark Pro

1 export interface RobotPose { robotId: string; xCm: number; yCm: number; headingDeg: number; seq: number; }

Why interviewers care

Warehouse Robotics interviews reward crisp scope, explicit trade-offs, and failure stories—not generic microservice diagrams.

Interview checkpoint

Name one failure story for Problem Statement: Autonomous Warehouse Robotics that proves you understand real outages, not happy-path diagrams.

Key Highlights

•Orchestrate hundreds of AMRs picking totes in a fulfillment grid
•Integrate WMS pick waves with real-time path reservations
•Guarantee human-safe motion with sub-second replanning
•Maximize picks-per-hour without aisle deadlocks

1	def cells_for_path(path: list[tuple[int, int]], t0_ms: int, speed_mps: float) -> list[tuple[tuple[int,int], int, int]]:
2	"""Return (cell, enter_ms, exit_ms) reservations along a polyline."""
3	out = []
4	t = t0_ms
5	for a, b in zip(path, path[1:]):
6	dt = int(1000 * grid_distance(a, b) / speed_mps)
7	out.append((b, t, t + dt))
8	t += dt
9	return out

Problem Statement: Autonomous Warehouse Robotics

Fleet charter (1)

Why interviewers care

Interview checkpoint

Key Highlights

Section Rescue Kit