Field evaluation and optimization of a lightweight autonomous lidar-based UAV system based on a rigorous experimental setup in boreal forest environments

Interest in utilizing autonomous uncrewed aerial vehicles (UAVs) for under-canopy forest remote sensing has increased in recent years, resulting in the publication of numerous autonomous flight algorithms in the scientific literature. To support the selection and development of such algorithms, a reliable comparison of existing approaches based on published studies is essential. However, reliable comparisons are currently challenging due to widely varying experimental setups and incomplete reporting practices. This study proposes a standardized experimental setup for evaluating autonomous under-canopy UAV systems to fill this gap. The proposed setup emphasizes quantitative reporting of forest complexity, visual representation of test environments, execution of multiple repeated flights, and reporting of flight success rates alongside qualitative flight results. In addition, flights at multiple target speeds are encouraged, with reporting of realized flight speed, mission completion time, and point-to-point flight distance. The proposed setup is demonstrated using a lightweight lidar-based quadrotor employing state-of-the-art open-source algorithms, evaluated through extensive experiments in two natural boreal forest environments. Based on a systematic evaluation of the original system, several improvements were introduced. The same experimental protocol was then repeated with the optimized system, resulting in a total of 93 real-world flights. The optimized system achieved success rates of 12/15 and 15/15 at target flight speeds of 1 m/s and 2 m/s, respectively, in a medium-difficulty forest, and 12/15 and 5/15 in a difficult forest. Adoption of the proposed experimental setup would facilitate the literature-based comparison of autonomous under-canopy flight systems and support systematic performance improvement of future UAV-based forest robotics solutions.

💡 Research Summary

The paper addresses a critical gap in the evaluation of autonomous UAV systems designed for under‑canopy forest remote sensing. While many recent studies have introduced novel flight algorithms, their experimental validation is often limited to a single flight, lacks quantitative description of the forest environment, and does not report success rates or failure causes in a reproducible manner. To remedy this, the authors propose a comprehensive, standardized experimental protocol that includes (1) quantitative forest‑complexity metrics (tree density, number of low‑hanging branches), (2) visual documentation of the test sites, (3) multiple repeated flights at several target speeds, (4) detailed logging of realized speed, mission completion time, and point‑to‑point distance, and (5) explicit reporting of flight success rates together with qualitative observations of failures and near‑misses.

The protocol is demonstrated on a lightweight quadrotor equipped with a 3‑D lidar, an IMU, and onboard computing. The software stack combines the IPC obstacle‑based path‑planning algorithm (Liu et al.) with the LT‑A‑OM lidar SLAM system (Zou et al.). An initial set of 33 flights was conducted in two boreal forest sites of differing difficulty (medium and difficult). Flights were performed at target speeds of 1 m/s and 2 m/s, and each flight’s actual speed, distance, and completion time were recorded. The original system achieved modest success rates (7/15 and 9/15 at 1 m/s and 2 m/s in the medium forest; 8/15 and 4/15 in the difficult forest). Failure analysis identified three primary causes: (i) insufficient lidar point density leading to missed obstacles, (ii) control‑loop latency that became critical at higher speeds, and (iii) SLAM drift caused by dense low‑hanging branches and foliage.

Based on this analysis, three key improvements were implemented: (a) increasing lidar scan rate from 10 Hz to 20 Hz to provide fresher obstacle data, (b) tuning the voxel‑grid filter to retain enough points while reducing noise, and (c) augmenting the path‑planning cost function with a speed‑dependent safety margin that penalizes aggressive maneuvers at higher velocities. After these modifications, an additional 60 flights were executed under the same protocol. In the medium‑complexity forest the optimized system achieved 100 % success at both 1 m/s and 2 m/s (15/15 each). In the difficult forest, success rates rose to 80 % (12/15) at 1 m/s and 33 % (5/15) at 2 m/s. The remaining failures at 2 m/s were still linked to rapid speed changes in densely branched sections, indicating that further work on high‑speed control robustness is needed.

The authors compare these results with those reported in the literature (e.g., Liu et al. 2024, Ren et al. 2025). Those studies often list target speeds and distances but omit forest‑complexity quantification and do not disclose the number of failed attempts, making direct performance comparison impossible. By contrast, the present work provides a full accounting of successes, failures, and environmental parameters, enabling an objective benchmark.

The discussion emphasizes that the proposed standardized protocol not only facilitates fair comparison across different algorithms and hardware configurations but also serves as a development tool for iterative improvement, as demonstrated by the clear performance gains after the three algorithmic refinements. The authors argue that adopting such a protocol will raise the reliability of UAV‑based forest sensing, accelerate the transition from laboratory prototypes to operational systems, and support meta‑analyses that can identify broader trends in algorithmic performance under varying forest conditions.

In conclusion, the paper delivers (i) a rigorously defined experimental framework for under‑canopy UAV testing, (ii) a concrete case study showing how systematic analysis and targeted algorithmic tweaks can substantially improve flight success, and (iii) a call for the community to adopt the protocol to enable reproducible, comparable, and scalable research in autonomous forest robotics. Future work is suggested in the areas of richer obstacle modeling (including foliage dynamics), high‑speed control stability, and multi‑UAV coordination, all evaluated under the same standardized metrics.