How Autonomous Delivery Robots Navigate Sidewalks at Night

Autonomous delivery robots use a combination of sensors, cameras, and pre-mapped data to move along sidewalks after dark. This guide explains the core navigation steps, the technology involved, and common pitfalls—so you can understand how these machines safely find their way when the sun goes down.

What You’ll Need

• A basic understanding of robotics or navigation concepts (not required, but helpful)
• Access to a sidewalk or urban environment where delivery robots operate (for observation)
• Patience—nighttime navigation involves many overlapping systems

Step-by-Step

1. Start with a pre-mapped route. Before the robot leaves the depot, its onboard computer loads a detailed map of the delivery area. This map includes sidewalk boundaries, crosswalks, curb ramps, and known obstacles like fire hydrants or benches. The robot uses this baseline to plan a path from point A to point B.
2. Activate the sensor suite. At night, the robot relies on a combination of sensors: LiDAR (light detection and ranging) sends out laser pulses to measure distances, cameras capture visual data, and ultrasonic sensors detect objects close to the ground. These work together to create a real-time 3D view of the surroundings.
3. Use infrared or low-light cameras. Standard cameras struggle in darkness, so many robots switch to infrared (IR) or low-light cameras. IR cameras illuminate the scene with invisible light, allowing the robot to see pedestrians, animals, and obstacles even in near-total darkness.
4. Fuse sensor data. The robot’s computer combines data from LiDAR, cameras, and ultrasonic sensors into a single “occupancy grid”—a digital map that marks which areas are free and which are blocked. This fusion reduces errors that any single sensor might have in low light.
5. Detect and classify objects. Using machine learning algorithms, the robot identifies objects in its path: a person, a bicycle, a trash can, or a curb. At night, this is harder because shapes are less distinct, so the robot may rely more on LiDAR’s point cloud to determine an object’s size and movement.
6. Navigate around obstacles. When the robot encounters an obstacle, it calculates a new path around it, staying within the sidewalk boundaries. It slows down, stops if needed, and then proceeds once the path is clear. At night, it may be more cautious, reducing speed to account for lower visibility.
7. Follow traffic rules and crosswalks. At intersections, the robot uses its cameras to detect traffic signals and crosswalk markings. It only crosses when the pedestrian signal is green and checks for oncoming vehicles using LiDAR and radar. Some robots have dedicated sensors that detect the glow of headlights to avoid collisions.
8. Handle dynamic obstacles. Moving objects like joggers or dogs are tracked continuously. The robot predicts their trajectory and adjusts its own path to maintain a safe distance. At night, it may use reflective tape or lights on the robot to make itself visible to humans and vehicles.
9. Use GPS and localization. While GPS is not accurate enough for sidewalk-level navigation, the robot combines GPS with its sensors and the pre-loaded map to determine its exact position. This process, called localization, ensures the robot knows which side of the street it’s on and which way it’s facing.
10. Communicate with a remote operator. If the robot encounters a situation it cannot resolve—like a closed sidewalk or a crowd—it sends a video feed to a remote human operator who can take over temporarily. This is more common at night when visibility is limited.
11. Arrive at the destination. Once the robot reaches the delivery point, it uses its cameras to locate the exact drop-off spot, often a front door or lobby. It then signals its arrival via lights or a message on its screen, and the user retrieves the package.

Common Mistakes

• Relying solely on GPS. GPS can be off by several feet, especially in urban canyons or under trees. Robots must fuse GPS with sensor data to stay on the sidewalk.
• Ignoring lighting changes. A robot that works well under streetlights may struggle in unlit alleys. Operators must ensure the robot’s sensors are calibrated for varying light conditions.
• Overlooking reflective surfaces. Wet pavement or glass storefronts can confuse LiDAR, causing false readings. Advanced robots filter out these reflections, but it’s a common source of navigation errors.
• Not accounting for seasonal changes. Leaves, snow, or ice can alter the sidewalk’s appearance and make it slippery. Robots must adapt their traction and speed accordingly, which some models fail to do.
• Assuming all obstacles are static. At night, people may be less predictable—walking erratically or appearing suddenly from doorways. Robots that don’t continuously update their obstacle tracking can collide.

Tips

• Use redundancy. Rely on multiple sensor types—LiDAR, cameras, ultrasonic—so that if one fails, others can compensate.
• Maintain a conservative speed. At night, slower speeds give the robot more time to react to unexpected obstacles. Many robots automatically reduce their max speed after dark.
• Add visible lighting. Equip the robot with bright LED lights or reflective strips so pedestrians and vehicles can see it from a distance. This reduces the chance of human-caused collisions.
• Update maps regularly. Sidewalks change—construction, new benches, or temporary closures. Regularly re-mapping the area ensures the robot’s baseline data is accurate.
• Test in various weather conditions. Rain, fog, or snow can scatter LiDAR and obscure cameras. Test the robot in these conditions to fine-tune its sensor fusion.

Troubleshooting

• Robot stops unexpectedly. This often means it detected an obstacle it cannot classify. Check the sensor logs—if the obstacle is a reflection or a shadow, adjust the sensor thresholds or add more training data for night scenes.
• Robot drifts off the sidewalk. If the robot veers onto the road, its localization may be off. Re-calibrate the LiDAR and cameras, and ensure the map is up-to-date. Also check for magnetic interference from nearby metal structures.
• Poor performance in heavy rain. Raindrops can reflect LiDAR and blur cameras. Switch to a sensor fusion mode that prioritizes radar or ultrasonic data, which are less affected by precipitation.
• Camera glare from streetlights. Bright lights can cause lens flare, blinding the camera. Use anti-glare coatings or adjust the camera’s exposure settings to handle high contrast.
• Robot gets stuck on a curb. If the robot can’t climb a curb, it may be a mechanical issue or the curb height exceeds the robot’s capability. Re-route around the curb or improve the robot’s suspension.
• Battery drains quickly at night. Night operation often requires more power for lights and sensors. Optimize power usage by dimming lights when not needed and using energy-efficient sensor modes.

Frequently Asked Questions

Do delivery robots use headlights?

Yes, most have LED lights that illuminate the path ahead and make the robot visible to others. These lights are usually always on, but may be brighter at night.

Can robots see in complete darkness?

Not completely. They rely on infrared cameras and LiDAR, which work without visible light, but they still need some ambient light or their own IR illumination to function optimally.

What happens if a robot encounters a closed sidewalk at night?

It will either find an alternative route on the map or contact a remote operator for assistance. It will not attempt to go onto the road unless it’s a designated crossing.

How do robots avoid hitting pedestrians at night?

They use a combination of LiDAR and cameras to detect people, even in low light. The robot slows down and gives pedestrians a wide berth, often stopping if a person is too close.

Are delivery robots allowed to operate at night?

Regulations vary by city and country. Some require special permits or restrict nighttime operations. Always check local laws before deploying robots.

Can robots navigate in fog or snow at night?

With difficulty. Fog scatters LiDAR and snow covers landmarks. Many robots are not fully reliable in these conditions and may pause operations until weather improves.

Related

• Portable Power Stations vs Gas Generators: Which Backup?
• Dinosaur Eggshells vs Modern Bird & Reptile Eggshells
• Repair Corrupted HEIC Files on Mac: Full Guide