← Amazon Interview Insights

Amazon·Software Engineer·Onsite - System Design / Architecture·Senior

Senior
May 2026

Summary

Amazon SWE system design round, one big open-ended question about robot navigation and human avoidance. No fluff, just a deep technical dive that took the whole session.

Questions Asked (1)

Q1

Design a mobile robot navigation system for an indoor/warehouse environment that can safely avoid humans while moving toward a goal. Cover sensor selection, core data structures, system architecture with module breakdown, planning strategy for dynamic obstacles, and write pseudocode for a main control loop.

System DesignAlgorithms & Data StructuresTechnical Trade-offs
Author's notes

This was a beast of a question.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying requirements (warehouse layout, human traffic, safety constraints) and then propose a layered architecture with sensors, perception, planning, and control. Emphasize safety through redundancy and dynamic replanning, and walk through a main control loop that integrates sensing, planning, and actuation.

Pro tip: Highlight the importance of fail-safe behaviors and graceful degradation—e.g., if a sensor fails, the robot should slow down or stop safely. Also, mention how you would validate the system through simulation and real-world testing with edge cases.

1. Clarify Requirements and Constraints

Ask about environment size, human density, speed requirements, safety standards, and payload. This shapes sensor and algorithm choices.

2. Select Sensors and Perception Stack

Choose a combination of sensors (e.g., LiDAR, depth cameras, ultrasonic) for 360° coverage and redundancy. Describe perception modules for human detection and tracking.

3. Design System Architecture and Data Structures

Outline modules: sensor interface, perception, world model (e.g., occupancy grid, costmap), planner, and controller. Explain key data structures like dynamic costmaps and human trajectory predictions.

4. Define Planning and Control Strategy

Use global planner (e.g., A*) for static map and local planner (e.g., DWA, TEB) for dynamic obstacles. Incorporate human motion prediction and social navigation norms.

5. Write Pseudocode for Main Control Loop

Present a loop that senses, updates world model, replans if needed, and sends velocity commands, with safety checks and fallback behaviors.

Key Points to Mention

  • Sensor fusion for robust human detection (e.g., LiDAR + camera + ultrasonic)
  • Dynamic costmap or velocity obstacle approach for real-time obstacle avoidance
  • Safety mechanisms: emergency stop, speed limiting near humans, fail-safe on sensor loss
  • Human trajectory prediction and social-aware navigation to avoid discomfort
  • Trade-offs between reactive and deliberative planning, and between optimality and real-time performance
  • Simulation and testing (e.g., Gazebo, ROS) for validation before deployment

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.