DJI Matrice 400 vs 350: LiDAR Efficiency Comparison

First, based on the specifications of both systems, the M400 when equipped with the L2 has a flight time of 59 minutes, which is 40% longer compared to the M350 or M300. This difference increases to 53 percent when we compare the estimated maximum effective flight time, which only accounts for the mission duration. When applying the estimated maximum effective flight time to the mission planner, we have a clear comparison in terms of operation site coverage at different flight altitudes. Overall, due to longer effective flight time, the M400 provides over 80 percent more coverage per flight when compared to the M350 or the M300.

 

*The maximum effective time refers solely to the mission duration between the first and last waypoints, excluding other operational phases such as UAV takeoff, transit to the first waypoint, return to the home point, and landing. These segments are considered non-effective mission time as they do not contribute directly to data acquisition. For comparison purposes between different systems, a standard deduction of 10 minutes is applied to account for this non-effective time. However, this value may vary depending on Return-to-Home (RTH) settings and can be further influenced by real-world factors such as wind conditions, terrain complexity, and weather. ** The mission was configured with 60% side overlap and 70% front overlap for LiDAR data collection. Efficiency mode was disabled to prioritize data quality, and IMU calibration was enabled. Elevation optimization was not applied. The flight boundary was defined using a square shape; however, due to the complexity of the actual polygonal boundary, the IMU calibration sequence may be extended. This is caused by the need for mid-route calibration and additional corrections during turns. Please note that real-world performance may be further influenced by environmental factors such as wind conditions, terrain variation, and weather.

 

In collaboration with the DJI certified surveying partner Cansel and CA5 Survey, we secured a comprehensive test site in Big Rock, Malibu, California. This 160-acre location was partially damaged during the 2025 California wildfires. The fires, which ravaged over 250,000 acres across Southern California, were among the most destructive in the state's recent history, fueled by extreme drought conditions and unusually strong Santa Ana winds. Several structures within our test site were destroyed, creating a real-world scenario requiring detailed reconstruction surveys for insurance claims and rebuilding efforts.

 

The site features varied terrain (including steep hillsides and flat areas), diverse vegetation conditions (ranging from completely burned to partially affected areas), and structural remains, for comparing drone performance in post-disaster recovery applications. For this assessment, we deployed the DJI L2 LiDAR sensor on both drone platforms to capture high-resolution ground data, enabling digital terrain following capabilities for the M350 and the more sophisticated real-time terrain following functionality available with the M400.

 

For the M400, the process works as follows:

 

First, we need to define the boundary of our scan. This boundary is determined and reviewed by the survey team to ensure it sufficiently covers the operation site. The boundary polygon can also be shared with the ground team for safety awareness, letting them know when and where the drone will be operating.

 

We then used DJI FlightHub 2 to create missions for the DJI M400, then synced them to the DJI Pilot 2 app via cloud sync, which synchronized the KMZ mission files. While we could have planned missions directly using the DJI Pilot 2 app on the remote controller, creating them on a desktop is more convenient. FlightHub 2 runs in a web browser environment, eliminating the need to handle the remote controller for mission planning. Additionally, since missions are stored in the cloud, multiple team members can access and modify them online, ensuring the mission is always up to date when needed. DJI FlightHub 2 does not support mission planning for the M350 platform. To work around this, we downloaded the M400 L2 mission in the Pilot 2 app (which had been synced from FlightHub 2), modified it within Pilot 2, and changed the platform from M400 to M350. This approach preserved all sensor and flight settings.

 

To ensure a fair comparison between the two platforms, we used identical payload settings, flight altitude (325ft AGL), flight speed (20 mph), overlap ratios (60% side and 70% forward), and flight boundaries.

 

Due to the site's significant terrain variations, we implemented terrain following for both the M350 and M400 flights to obtain uniform Ground Sampling Distance (GSD) and consistent LiDAR point density throughout the entire survey area; however, each platform used a different approach:

 

The M400 features real-time terrain following for both mapping missions and manual flights. This is achieved through its more advanced visual sensors. We only need to set the above ground level elevation under the mission planning with in FlightHub 2 and enable the Real Time Terrain Follow option, and the system automatically maintains consistent altitude over varying terrain during the flight.

 

 

For the M350, which lacks the Real-Time Terrain Follow feature available on the M400, we needed to apply a terrain or surface model in the Pilot 2 mission planning interface. This required obtaining terrain data of the site to generate the terrain-follow route. In the mission planning settings, we first selected "AGL" (Above Ground Level) as the altitude mode, then selected "DSM Files" to add the terrain or surface model in the Pilot 2 app to serve as reference terrain.

 

There are two options available. The first, "Import Local File," requires importing a DSM/DEM file in geotiff format with WGS84 and Ellipsoidal height system in meter units. The second option, "Download from Internet," allows the Pilot 2 app to directly download the ASTER GDEM V3 30-meter resolution model based on the mission boundary. Both the M350 and M400 support DSM/DEM file import and ASTER download for terrain following.

 

While the second option is more convenient, ASTER GDEM often fails to account for real-time terrain changes due to outdated data, and its 30m resolution may not accurately represent terrain variations. To ensure optimal results with uniform point density and ground sampling distance, we processed high-resolution DSM from previously captured M400 L2 photogrammetry data. We processed this dataset with low-resolution output settings on-site using DJI Terra software for optimal processing speed on site, once done, we then selected "Import Local Files" to load the transferred DSM file for mission planning.

 

During this mission planning phase, we noticed that the workflow is significantly simplified with the M400's real-time terrain following capability. The M350's requirement for DSM/DEM files adds extra work and introduces potential human operation error risks related to file format, coordinate system, and import size limitations.

 

We could have used the ASTER GDEM to simplify this process; however, using potentially outdated terrain models creates uncertainty for operations in sites with frequent terrain changes, such as mining and construction scenarios. Therefore, importing processed DSM is a more reliable approach for these situations.

 

After creating both missions, we immediately observed an 8-minute difference in flight duration—an approximate 18% improvement in efficiency with the M400 compared to the M350. Since both flights used the same speed, altitude, and mission parameters, we initially attributed the discrepancy to differences in terrain following behavior. The M400 uses real-time terrain following, while the M350 calculates elevation changes in advance based on the imported terrain model.

 

Our initial hypothesis was that the M350's preplanned terrain-following route would more accurately represent real-world flight saturation, while the M400’s route might appear artificially shorter due to not accounting for actual elevation changes during mission planning. We suspected that the 8-minute gap would eventually be filled during the M400's flight as it ascended and descended in response to terrain in real time.

 

However, our results disproved that assumption. In practice, the M400’s real-time terrain following is precisely what enabled the 16% improvement in flight efficiency.

 

The survey technician with CA 5 Survey first set up 23 control targets across the operation site with recorded state projection coordinates. These targets serve to validate the UAV system's precision and accuracy, and can also constrain the model to improve absolute accuracy. Each control point was placed on a GCP target with reflective markings to make them easier to identify in point cloud reflectivity views.

 

The DJI D-RTK 3 was utilized as the correction reference station during data collection. RTCM data recorded from the base was later submitted to OPUS to obtain a more precise solution for the D-RTK 3's APC (Antenna Phase Center) coordinates. These refined coordinates were then used to update the original base station position in DJI Terra. The updated base position was subsequently applied in both the PPK processing of M350 RTK with L2 data and the RTK processing of M400 RTK with L2 data, ensuring improved geolocation accuracy across both datasets.

7. Conclusion 

Special thanks to Cansel and to Chris Nelson, PLS at CA 5 Survey, for their collaboration in organizing and executing this efficiency comparison test. Our testing and comparative analysis show that the M400 platform offers a clear efficiency advantage over the M350 when performing LiDAR mapping at midsize operational sites. This advantage stems from key platform improvements, including extended flight endurance, enhanced real-time terrain following, and streamlined operational complexity. These capabilities collectively result in significantly higher coverage efficiency per flight, reducing the need for frequent battery swaps—particularly on larger sites—without compromising data quality.

 

Under identical conditions, the M400 demonstrated a 16% improvement in overall flight efficiency and achieved approximately 39% greater single-flight coverage compared to the M350.