Buyer’s guide

Best Sensors for Autonomous Navigation

The best sensors for autonomous navigation are a combination, not a single device. Most autonomous drones pair an IMU with RTK GNSS outdoors, add cameras for visual-inertial odometry when GPS drops out, and use lidar, radar, time-of-flight or optical flow sensors to see obstacles and hold position near the ground.

This guide covers each sensor type with a current product example, so you can match sensors to your aircraft and environment. Specifications are the manufacturers’ own.

Method

How we chose

01One example per sensor typeEach entry names a current product.
02Proven in autonomyWe favored sensors with PX4, ArduPilot or ROS 2 support.
03Published specificationsEach entry uses figures from the maker’s site or datasheet.
At a glance

Quick comparison

Sensor (example) HQ Best for Notable
Stereo VIO camera (Stereolabs ZED X Mini) San Francisco, CA GPS-denied position tracking Dual global-shutter sensors, IMU, IP67
Active stereo depth (RealSense D455, D435i) Santa Clara, CA Short-range obstacle depth Global shutter, built-in IMU
3D lidar (Ouster OS0, OS1) San Francisco, CA Mapping and dense 3D perception Rev8 OS1 Max: 256 channels, native color
Scanning lidar rangefinder (LightWare SF45/B) Boulder, CO Lightweight collision prevention 59 g, up to 50 m
Radar (Echodyne EchoFlight) Kirkland, WA Airborne detect and avoid Under 1 kg, under 50 W
Time-of-flight (ST VL53L5CX) Geneva, Switzerland Close-range multizone ranging 8×8 zones, up to 4 m
Optical flow (ARK Flow) Salt Lake City, UT Low-altitude position hold Flow, 30 m distance sensor and IMU
RTK GNSS (u-blox ZED-X20P, Septentrio mosaic-X5) Thalwil, Switzerland; Leuven, Belgium Centimeter-level outdoor position All-band RTK
IMU and INS (VectorNav VN-100, VN-300) Dallas, TX Attitude and heading VN-300 dual-antenna GNSS heading
Thermal (Teledyne FLIR Boson+) Goleta, CA (OEM operations) Night and low-visibility perception 640×512, 20 mK or better
01

Stereo VIO camera: Stereolabs ZED X Mini

Best for
Tracking position without GPS using cameras and inertial data.

The ZED X Mini pairs two 1920×1200 global-shutter color sensors with a 16-bit IMU for depth sensing and visual-inertial tracking. Global shutters avoid the skew that rolling shutters show on a moving aircraft. It connects over GMSL2 with hardware sync for multi-camera rigs and has an IP67 aluminum enclosure. Stereolabs says its Neural Depth Engine produces depth from 10 cm to tens of meters. Ouster acquired Stereolabs in February 2026, and Stereolabs now operates as a wholly owned subsidiary with its co-founders still in leadership.

02

Active stereo depth: RealSense D455 and D435i

Best for
Short-range obstacle depth on robots and slow drones.

The RealSense D455 has global-shutter depth and RGB sensors, a Bosch BMI055 IMU and an ideal range of 0.6 to 6 m, with under 2% depth error at 4 m. The D435i adds an IMU to the compact D435. RealSense sells these cameras under license from Intel. RealSense spun out of Intel in July 2025, and on 22 September 2026 Cognex announced an agreement to acquire it, with closing expected in the fourth quarter of 2026. Buyers should watch for any product-line changes after the deal.

03

3D lidar: Ouster OS0 and OS1

Best for
Dense 3D mapping and perception on mid-size and larger platforms.

Ouster makes digital lidar for robotics, drones, mapping and defense. The ultra-wide OS0 targets short-range, high-precision perception, and Flyability is evaluating the Rev8 OS0 for 3D mapping in GPS-denied industrial sites. The Rev8 OS1 Max, aimed at aerial mapping, has 256 channels with native color, ±0.25 cm precision and detection of 10%-reflective targets up to 200 m, per Ouster. Lidar measures distance directly, so it works at night and on plain surfaces that confuse cameras, at the cost of more weight and power.

04

Scanning lidar rangefinder: LightWare SF45/B

Best for
Lightweight collision prevention without a companion computer.

The SF45/B is a 59 g scanning microLiDAR with a range of 0.2 to 50 m and an adjustable field of view up to 320°. It takes up to 5,000 readings per second and up to 5 sweeps per second, and LightWare says it tolerates changing light, wind and noise. Both PX4 and ArduPilot support it for collision prevention, so it can protect a small drone without extra compute. LightWare is headquartered in Boulder, Colorado, with manufacturing in South Africa.

05

Radar: Echodyne EchoFlight

Best for
Detecting other aircraft for detect-and-avoid.

EchoFlight is an airborne electronically scanned array radar that weighs under 1 kg and uses under 50 W. It tracks in range, azimuth, elevation and Doppler, with typical tracking ranges from about 500 m to 2,750 m depending on target size, and uses machine learning for classification. Radar works through darkness, fog and dust where cameras struggle. At chip level, TI’s IWR6843AOP, a 60 to 64 GHz mmWave sensor with its antenna on the package, is a smaller option for short-range sensing.

06

Time-of-flight: ST VL53L5CX

Best for
Close-range obstacle sensing in tight spaces.

The VL53L5CX is a multizone direct time-of-flight sensor that measures up to 8×8 zones across a 65° diagonal field of view, with ranging up to 4 m at up to 60 Hz. ST’s histogram processing lets it detect several objects in the field of view and measure distance regardless of target color. That makes it a basic low-resolution depth sensor for indoor drones, landing checks and close-quarters obstacle detection, at a fraction of lidar’s size and cost. The field of view can also be narrowed in software.

07

Optical flow: ARK Flow

Best for
Holding position at low altitude when GPS is weak.

ARK Flow is an open-source DroneCAN sensor that combines a PAW3902 optical flow sensor, a Broadcom AFBR-S50 distance sensor with up to 30 m range and an ICM-42688-P IMU. The flow sensor tracks from 80 mm to 30 m and works in light as low as 9 lux. The distance sensor handles up to 200k lux of ambient light. PX4 and ArduPilot both support ARK Flow, and ARK Electronics designs and manufactures it in the USA. It is a common first step for GPS-denied hover.

08

RTK GNSS: u-blox ZED-X20P and Septentrio mosaic-X5

Best for
Centimeter-level positioning outdoors.

The u-blox ZED-X20P is an all-band GNSS module with RTK, PPP-RTK and PPP support in the same footprint as the popular dual-band ZED-F9P. Septentrio’s mosaic-X5 is a multi-band module with a 100 Hz update rate and AIM+ interference mitigation, with a Premium version for stronger anti-jamming and anti-spoofing. Hexagon completed its acquisition of Septentrio in March 2025, and Advent International completed its acquisition of u-blox in November 2025. RTK needs a correction source, either a local base station or a network service, so plan for that in operations.

09

IMU and INS: VectorNav VN-100 and VN-300

Best for
Reliable attitude and heading, including in low-dynamics flight.

The VN-100 is an IMU and AHRS that combines 3-axis accelerometers, gyros and magnetometers with a barometer and a 32-bit processor, giving calibrated IMU data and a real-time attitude solution. The VN-300 adds two GNSS receivers and antennas, so it can measure heading without relying on vehicle motion or magnetometers, which helps hovering multicopters that cannot rely on motion for heading. PX4 supports VectorNav units. VectorNav, based in Dallas, Texas, has supplied navigation systems to military, aerospace, marine and robotics integrators since 2008.

10

Thermal: Teledyne FLIR Boson+

Best for
Seeing people, vehicles and heat sources at night or in smoke.

Boson+ is a 640×512 longwave infrared module with a 12-micron pixel pitch and thermal sensitivity of 20 mK or better. It sees through darkness, smoke and most fog, and it works with Teledyne FLIR’s Prism AI software for detection and tracking. Teledyne FLIR also sells the Hadron 640R+, which pairs a radiometric Boson core with a 64 MP visible camera in 56 grams. Thermal is one of the categories in the FCC’s proposed limits on foreign components, so origin matters for U.S. programs.

FAQ

Common questions

What sensors does a drone need to fly autonomously?

At minimum, an IMU, a barometer and GNSS for outdoor flight. For autonomy beyond waypoints, add cameras for visual-inertial odometry and at least one ranging sensor, such as lidar, radar, time-of-flight or optical flow with a distance sensor, to detect obstacles and hold position near the ground.

Lidar or cameras for obstacle avoidance?

Cameras are light, cheap and give rich detail, but struggle in darkness and on plain surfaces. Lidar measures distance directly and works at night, but adds weight and cost. Many autonomous drones use cameras for navigation and a small lidar or time-of-flight sensor as a safety layer.

How do drones navigate without GPS?

Most use visual-inertial odometry, which fuses camera images with IMU data to track motion. Optical flow with a downward distance sensor helps at low altitude, and lidar SLAM works well in structured spaces. Accuracy depends on light, texture and speed, so test in your real environment.

Do sensor choices affect NDAA or FCC status?

They can. NDAA sourcing rules cover specific component types and manufacturers, and the FCC has proposed further limits on foreign thermal and lidar components. These are separate reviews. Confirm each sensor’s origin and the finished product’s status with your supplier and compliance team.

See all 310 questions in the drone and robotics FAQ →Look up terms in the glossary →

Integrating the sensor suite

Sensors must be mounted, synchronized and fused correctly. Aerora offers payload and sensor integration as part of its ODM services. For partners who build complete autonomy hardware, see the best robotic autonomy hardware partners.

Related reading: Best drones for infrastructure inspection.

Sources (38)
Start a program

Ready to build what’s next?

Whether you’re developing autonomous aircraft, robotics, intelligent sensing systems or connected platforms, Aerora can help bring your vision to life.

Submit a design inquiry