Flying through cluttered and dynamic environments with LiDAR

Fangcheng Zhu; Fu Zhang; Hairuo Wei; Haotian Li; Huajie Wu; Wenyi Liu; Yunfan Ren; Zheng Liu

REVIEW 3 major objections 4 minor 22 references

Flying through cluttered and dynamic environments with LiDAR

T0 review · 3 major / 4 minor · reviewed 2026-08-16 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A LiDAR-only system can fly a drone through cluttered forests while avoiding fast-moving obstacles onboard.

desk verdict A credible systems integration whose key new component (DynIPC) is never isolated in the evaluation—the predictive-avoidance claim needs an ablation and prediction-error statistics before it lands. read the letter →

arxiv 2504.17569 v1 pith:TX7D53R3 submitted 2025-04-24 cs.RO cs.SYeess.SY

Huajie Wu , Wenyi Liu , Yunfan Ren , Zheng Liu , Hairuo Wei , Fangcheng Zhu , Haotian Li , Fu Zhang This is my paper · ORCID

classification cs.ROcs.SYeess.SY

keywords LiDAR-basedUAVnavigationdynamicobstacleavoidanceintegratedplanningandcontrolmovingobjectdetectiontrajectorypredictionoccupancygridmapclutteredenvironmentsreal-timeonboardflight

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper makes the case that a single LiDAR sensor, an onboard computer, and the right division of labor between perception and planning are enough for a drone to fly through cluttered, dynamic surroundings. The system detects moving points without relying on object classes or colors, predicts where each tracked obstacle will be over a two-second horizon, and folds those predictions into a 100 Hz integrated planner-controller so it can dodge fast or suddenly appearing objects. In simulation it reaches success rates of 0.97 and 1.00 in two scenes, beating two open-source baselines, and in a forest it avoided thrown balls, pendulums, cars, and pedestrians with a measured perception-to-control latency of 11.54 ms. If the claims hold, LiDAR-only drones can operate safely in environments that previously required cameras, trained detectors, or reactive-only avoidance.

What carries the argument

The load-bearing mechanism is DynIPC, an integrated planning-and-control formulation in which the predicted position of each tracked obstacle at every MPC step enters the cost as $\max(d_{\mathrm{dyn}}^2 - \|p_n - p_{\mathrm{obs},n,k}\|^2_{R_o}, 0)$, simplified to a negative quadratic so the whole problem stays a convex QP solvable at 100 Hz. Around it sit two supporting mechanisms: a two-layer occupancy map (static points only for the static layer, predicted obstacle footprints for the dynamic layer, with velocity-scaled inflation $d_{\mathrm{inf}} = d_{\mathrm{rad}} + k\cdot\mathrm{vel}$) and the M-detector frame-out pipeline that supplies the moving points and trajectory fits. Together they let the planner treat static clutter as hard constraints and dynamic obstacles as soft, smoothly avoiding many moving objects without resorting to nonlinear optimization.

What would settle it

A controlled flight test with an obstacle that accelerates or turns sharply inside the two-second prediction horizon, in otherwise identical forest conditions, would settle the matter: if near-miss distances stay the same, the reactive replanner is doing the safety work; if collisions or unsafe clearances appear, the polynomial motion assumption is the load-bearing premise.

Watch

Extended reading notes

Core claim

The central claim is that dynamic obstacle avoidance can be solved onboard with LiDAR alone by separating the world into a static occupancy layer built only from static points and a dynamic layer built from predicted future positions of moving objects. Moving points are extracted by M-detector's frame-out mode, clustered with DBSCAN, associated across frames with the Hungarian algorithm, and fit to one of two motion models—constant velocity or free fall—to forecast positions up to two seconds ahead. The planner, DynIPC, extends the authors' earlier integrated planning and control work so that these predicted positions enter a convex MPC as soft quadratic penalties, while static obstacles remain hard corridor constraints; the same algorithm also searches the reference path with A* over grids that are both statically and dynamically safe. The paper's own summary of the result is that with low-latency perception and prediction-aware planning, the UAV avoids heterogeneous moving obstacles in cluttered real environments while all computation runs onboard.

Load-bearing premise

Predictive avoidance rests on the assumption that every tracked obstacle keeps moving with constant velocity or in free fall for the next two seconds; an obstacle that turns, brakes, or is pushed will not be where the map and cost function expect it to be.

Editorial extensions

If this is right

A LiDAR-only sensing suite, without cameras or pretrained object classes, is sufficient for cluttered dynamic navigation outdoors.
Fast, sudden-appearing obstacles can be handled at up to 8.75 m/s relative speed because the full perception-to-control chain runs in about 11 ms.
Because dynamic avoidance is a soft quadratic cost in a convex QP, many tracked obstacles can be considered simultaneously without breaking the 100 Hz control rate.
Using only static points to build safe flight corridors enlarges the navigable space and prevents ghost points from moving objects from blocking the path.
In the paper's comparisons, prediction-aware planning reaches the goal faster and at higher speed than reactive or learning-based baselines while maintaining a higher success rate.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The two-motion-model prediction (constant velocity or free fall) is the part most likely to limit real generality; a drone, car, or animal that turns or brakes within the two-second horizon would invalidate the predicted positions, and the system would then depend on the reactive replanning loop for safety.
The same static/dynamic map split and soft-constraint MPC could be transferred to ground robots or manipulation with point-cloud sensing, since nothing in the formulation is quadrotor-specific except the kinodynamic limits.
One testable extension is to replace the fixed velocity-scaled inflation with uncertainty-aware inflation derived from prediction covariance, which would make the safety margin adapt to how confident the tracker is.
The simulation comparison used a depth camera in all three systems to isolate the planning contribution; real-world results use LiDAR, so an end-to-end LiDAR simulation benchmark with the same baselines would directly test the sensor advantage.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Punchline: this is a competent systems paper, and the real-world flights are credible, but the headline claim that DynIPC's predictive avoidance is what makes it work is not supported by the evidence. The evaluation never isolates the contribution.

The genuinely new bit is DynIPC—extending the authors' earlier IPC planner with a soft cost for predicted positions of dynamic obstacles, plus static-only SFC and a dynamic-aware A* search. Putting M-detector, the two-layer ROG-Map, and DynIPC together into one onboard LiDAR stack is a useful engineering contribution. The real data are meaningful: 11.54 ms perception-to-control latency, flights through a forest with pendulums, balls, a car, and pedestrians. That is not nothing.

The weak link is exactly the trajectory-prediction module. Section IV-B fits only constant-velocity or free-fall polynomials over a 2 s window. The simulation obstacles are in reciprocating motion (6 and 2 m/s); a reciprocating trajectory reverses direction faster than the prediction horizon, so a sliding-window polynomial fit is systematically wrong around the turn. Yet the paper never reports prediction error, never ablate the prediction module, and never compares DynIPC against its own reactive IPC baseline with the same stack. The success rates (0.97 and 1.00) and the latency figure therefore do not establish that predictive avoidance is doing the safety work; the 100 Hz reactive replanning loop could be. This is a load-bearing gap in the paper's scientific claim, though not in the system demonstration.

Other soft spots, mostly minor: the simulation uses a depth camera instead of LiDAR for all methods, so the "LiDAR-based system" claim is only directly tested in the single real-flight sequence. Table I reports no variance over the 30 runs. The convexity assertion—setting Ro < Rp gives a positive semidefinite Hessian—is stated without proof; it may be true but it is not obvious from Eq. (2) as written. The citation pattern is heavily self-referential, but for an integration paper that is appropriate; the prior work is published and appears real.

Bottom line: this is a worthwhile systems paper for the UAV navigation community, and it deserves referee time. But the referee should ask for an ablation with the dynamic cost removed and for prediction-error statistics before the DynIPC contribution is accepted. If those come back, the paper becomes a solid reference architecture.

Referee Report

3 major / 4 minor

Summary. The paper proposes a complete LiDAR-based UAV navigation system for cluttered and dynamic environments. The perception module uses the authors' prior M-detector to classify moving points, clusters and tracks dynamic objects with DBSCAN and the Hungarian algorithm, and predicts their future positions over a 2 s horizon using either constant-velocity or free-fall polynomial fits. The planning module extends the authors' prior IPC framework into DynIPC by adding dynamic-obstacle predictions to a two-layer ROG-Map (static and dynamic layers) and to an MPC-based integrated planning-and-control problem, solved at 100 Hz. The system is evaluated in two Gazebo simulation scenes against Panther and Chen's work, where it reports higher success rates (0.97 and 1.00), lower flight times, and higher maximum velocities, and in a real-world forest experiment with a LIVOX Mid-360 LiDAR, where it reports a perception-to-control latency of 11.54 ms and successful avoidance of cars, balls, pedestrians, and pendulums.

Significance. If the claims hold, this would be a useful system-level demonstration: a single LiDAR-based onboard stack that detects and avoids heterogeneous moving obstacles in real time. The paper's strengths include the integration of previously developed components into a complete pipeline, the use of two open-source baselines for comparison, and a real-world test with diverse dynamic obstacles. The low reported latency and the high simulation success rates indicate practical potential. However, the evidence as presented does not yet establish the central contribution of predictive avoidance: the simulation substitutes a depth camera for the LiDAR sensor, the trajectory prediction model is not validated against the actual obstacle motions used in the tests, and the real-world validation is a single qualitative flight. The paper is clearly written and the supplementary video is a positive asset.

major comments (3)

[VII-A] The simulation experiments replace the LiDAR sensor with a depth camera for all methods, including the proposed system, and use ground-truth localization. This means the simulation results do not exercise the LiDAR-based perception module (M-detector), the LiDAR odometry, or the sensor fusion pipeline that the paper's title and abstract claim to validate. The real-world experiment in Section VII-B does use the LIVOX Mid-360, but it is a single qualitative demonstration without repeated trials or quantitative success statistics. To support the central claim of a LiDAR-based system, the authors should either run the simulation with a LiDAR sensor model (even at the cost of benchmark comparability) or explicitly reposition the real-world experiment as the primary validation of the LiDAR pipeline and provide repeated-trial statistics.
[IV-B, VII-A] The trajectory prediction module assumes that each tracked obstacle follows either constant-velocity or free-fall motion over a 2 s window, yet the simulation obstacles are described as being in reciprocating motion at 6 m/s and 2 m/s, and the real-world experiment includes a pendulum. A reciprocating or pendulum trajectory reverses direction within the prediction/MPC horizon, so the fitted linear or quadratic curve will place the predicted obstacle on the wrong side of the reversal. Both the dynamic occupancy layer (Section V) and the avoidance cost in Eq. (2) are built from these predicted positions. The paper reports no prediction-error statistics, no ablation comparing DynIPC with the reactive IPC baseline, and no analysis of cases where the prediction assumption failed. These omissions leave the central predictive-avoidance contribution unvalidated, because the 100 Hz reactive replanning loop described in Section VI-D could itself account for the reported success rates.
[Table I, VII-B] The simulation results in Table I report point estimates over 30 runs without variance, confidence intervals, or statistical tests, and the experimental setup does not state what randomness, if any, differentiates the runs. With success rates of 0.60 versus 0.97 and 0.07 versus 1.00, the claim that the system outperforms the baselines needs at least standard deviations or a description of the stochastic variations. In addition, the real-world section reports a single flight; the 11.54 ms latency and the 4.55 m/s maximum velocity are single measurements with no repeated trials. The empirical evidence is suggestive but not yet convincing at the level of a journal publication.

minor comments (4)

[VI-C, Eq. (2)] The simplification from Eq. (1a) to Eq. (2) pre-selects the obstacles based on the reference trajectory position pref,n rather than the planned position pn. This should be stated more explicitly, because if the planned trajectory deviates from the reference path, the set of active obstacle costs can be wrong; the current wording in the text describes the reference position as "the estimated optimal UAV position" without justifying that approximation.
[VII-A] The paper says M-detector is employed in the simulation, but M-detector was originally developed and validated for LiDAR point streams. The authors should clarify whether M-detector requires any parameter changes or recalibration when applied to depth-camera point clouds, since the simulation uses a depth camera for all methods.
[Table II] Table II lists vi,max = 4 as the maximum velocity in the x, y, z direction, but Table I reports maximum velocities of 4.87 m/s and 4.56 m/s. If vmax in Table I is the Euclidean norm of the velocity, this is consistent, but the distinction between per-axis limits and the reported norm should be stated explicitly to avoid confusion.
[General] The term "integrated planning and control" is used both for the overall system and for the specific MPC module (DynIPC). The paper would benefit from a consistent terminology, e.g., reserving "integrated planning and control" for the MPC-based back end and using "system" for the complete pipeline.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: prediction-based avoidance is extrapolation from fitted obstacle motion, not a fit to the claimed outcome, and the system is validated against external baselines.

full rationale

Walking the derivation chain: Section IV-B fits obstacle motion from observed positions using linear or quadratic polynomials; that fitted trajectory is then used to populate the dynamic occupancy layer (Section V) and the avoidance cost in Eq. (1a)/(2). This is extrapolation from measured history, not a parameter fitted to make the avoidance outcome come out. The MPC cost penalizes distance to predicted obstacle positions; the simplification that uses the reference position to select which obstacles to include is a heuristic but does not redefine the prediction as the planning output. The use of M-detector [13] and IPC [14] is component reuse supported by the cited prior publications and by this paper's own external comparison against Panther and Chen's method; no uniqueness theorem or hidden ansatz is imported from those citations. The two-model prediction assumption (constant velocity or free fall) and the absence of prediction-error statistics are validation limitations, not circularity: the reported success rates and 11.54 ms latency are measured outcomes, not consequences of a fit. No step in the paper reduces by construction to its own inputs, so no circular step is identified.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The system is an integration of previously published components, with hand-tuned weights and prediction model assumptions as the main free inputs. The central claim depends on the reliability of M-detector, the validity of the polynomial motion model, and a series of ad hoc cost and map simplifications that are not independently validated in this paper.

free parameters (5)

MPC weights Rp, Ru, Rc, Ro, Rv,N, Ra,N = Rp=200, Ro=20, Ru=Rc=0, Rv,N=Ra,N=30
Hand-tuned weights in cost (1); Ro < Rp is needed for the claimed convexity of (2), so the value is load-bearing for the solver to work.
d_dyn minimum safe distance = 0.8 m
Configured safety distance for dynamic obstacles in cost (1); chosen by hand, affects when the avoidance cost is active.
k velocity inflation weight = 0.01
Used in d_inf = d_rad + k*vel for dynamic layer inflation; hand-selected and tuned.
MPC horizon N and time step delta_t = N=15, delta_t=0.1 s
Discretization choices that define the 1.5 s prediction horizon of the controller.
Trajectory prediction window and object removal time = 2 s prediction, 1.0 s removal
Choice of prediction length and removal timeout affects the dynamic layer of the map and avoidance behavior.

assumptions (5)

domain assumption M-detector reliably classifies points as static or dynamic across varied object sizes, colors, and types.
The entire perception module is taken from prior work [13]; the present paper does not re-derive or benchmark its failure modes. Central to the full system claim.
domain assumption Dynamic obstacle motion over the next 2 s is well approximated by constant velocity or free fall.
Sec IV-B: trajectory prediction chooses between linear and quadratic fits with quadratic coefficient near g/2. No validation of prediction error is given.
domain assumption Reference positions from A* are sufficiently close to the optimal UAV positions that using them to select and evaluate dynamic obstacle cost is valid.
Sec VI-C: 'the UAV reference position pref,n is treated as the estimated optimal UAV position' before optimization. This decoupling makes the QP easy but is not proven.
ad hoc to paper The quadratic cost in Eq. (2) is positive semidefinite when Ro < Rp.
Stated in Sec VI-C as fulfilled by setting Ro smaller than Rp; with multiple dynamic obstacles per step the Hessian also depends on Km and obstacle geometry, so the condition is not sufficient in general.
domain assumption The two-layer occupancy grid map correctly represents free space from 30 Hz scans.
Map structure from ROG-Map [19]; the dynamic layer is binary and cleared each prediction update, assuming no stale or incorrect labels cause false collisions.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Flying through cluttered and dynamic environments with LiDAR." pith.science (2026). https://pith.science/paper/TX7D53R3

@misc{pith2026250417569,
  author       = {Pith},
  title        = {Pith review of: Flying through cluttered and dynamic environments with LiDAR},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TX7D53R3}},
  note         = {Machine review of arXiv:2504.17569}
}

read the original abstract

Navigating unmanned aerial vehicles (UAVs) through cluttered and dynamic environments remains a significant challenge, particularly when dealing with fast-moving or sudden-appearing obstacles. This paper introduces a complete LiDAR-based system designed to enable UAVs to avoid various moving obstacles in complex environments. Benefiting the high computational efficiency of perception and planning, the system can operate in real time using onboard computing resources with low latency. For dynamic environment perception, we have integrated our previous work, M-detector, into the system. M-detector ensures that moving objects of different sizes, colors, and types are reliably detected. For dynamic environment planning, we incorporate dynamic object predictions into the integrated planning and control (IPC) framework, namely DynIPC. This integration allows the UAV to utilize predictions about dynamic obstacles to effectively evade them. We validate our proposed system through both simulations and real-world experiments. In simulation tests, our system outperforms state-of-the-art baselines across several metrics, including success rate, time consumption, average flight time, and maximum velocity. In real-world trials, our system successfully navigates through forests, avoiding moving obstacles along its path.

Figures

Figures reproduced from arXiv: 2504.17569 by the authors.

**Figure 1.** UAV Navigating through a Cluttered and Dynamic Environment. A pendulum made of a tennis ball is introduced as a moving obstacle along the UAV’s path. The red circle highlights the enlarged obstacle, while the red arrow indicates its direction of movement at each time point. The blue box represents the UAV’s position. (a) The pendulum swung to the right. The UAV detected a potential collision and planned to move to … view at source ↗

**Figure 2.** The overview of our system. in local minima. Besides, if there is a narrow passage, the repulsive forces from obstacles on both sides can overpower the attractive force towards the goal, making it difficult or impossible for the UAV to pass through it. Overall, this method can not work well in complicated environments. Optimization-based methods have also been explored for dynamic environment planning [5]–[8], [12].… view at source ↗

**Figure 3.** Reference path searching based on the local map in a simplified [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Figures from the paper (3 more)

**Figure 4.** Figure 4: Simulation environments for UAV to fly through the dynamic [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]

**Figure 5.** Figure 5: The platform used in the real-world experiments. [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]

**Figure 6.** Figure 6: The real-world experiment performed in a cluttered and dynamic [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

22 extracted references · 17 canonical work pages

[1]

Adoption of unmanned aerial vehicle (uav) imagery in agricultural management: A systematic literature review,

M. A. Istiak, M. M. Syeed, M. S. Hossain, M. F. Uddin, M. Hasan, R. H. Khan, and N. S. Azad, “Adoption of unmanned aerial vehicle (uav) imagery in agricultural management: A systematic literature review,” Ecological Informatics , p. 102305, 2023

work page 2023
[2]

Airborne! uav delivery of blood products and medical logistics for combat zones,

D. T. Lammers, J. M. Williams, J. R. Conner, E. Baird, O. Rokayak, J. M. McClellan, J. R. Bingham, R. Betzold, and M. J. Eckert, “Airborne! uav delivery of blood products and medical logistics for combat zones,” Transfusion, vol. 63, pp. S96–S104, 2023

work page 2023
[3]

LiDAR-based Quadrotor for Slope Inspection in Dense Vegetation

W. Liu, Y . Ren, R. Guo, V . W. Kong, A. S. Hung, F. Zhu, Y . Cai, Y . Zou, and F. Zhang, “Lidar-based quadrotor for slope inspection in dense vegetation,” arXiv preprint arXiv:2409.13985 , 2024

work page Pith review arXiv 2024
[4]

Unmanned aerial vehicles for search and rescue: A survey,

M. Lyu, Y . Zhao, C. Huang, and H. Huang, “Unmanned aerial vehicles for search and rescue: A survey,” Remote Sensing , vol. 15, no. 13, p. 3266, 2023

2023
[5]

Panther: Perception-aware trajectory planner in dynamic environments,

J. Tordesillas and J. P. How, “Panther: Perception-aware trajectory planner in dynamic environments,” IEEE Access , vol. 10, pp. 22 662– 22 677, 2022

2022
[6]

Perception and avoidance of multiple small fast moving objects for quadrotors with only low-cost rgbd camera,

M. Lu, H. Chen, and P. Lu, “Perception and avoidance of multiple small fast moving objects for quadrotors with only low-cost rgbd camera,” IEEE Robotics and Automation Letters , vol. 7, no. 4, pp. 11 657–11 664, 2022

work page 2022
[7]

Flying in dynamic scenes with multitarget velocimetry and perception-enhanced planning,

H. Chen, C.-Y . Wen, F. Gao, and P. Lu, “Flying in dynamic scenes with multitarget velocimetry and perception-enhanced planning,”IEEE/ASME Transactions on Mechatronics , 2023

work page 2023
[8]

Vision-aided uav navigation and dynamic obstacle avoidance using gradient-based b- spline trajectory optimization,

Z. Xu, Y . Xiu, X. Zhan, B. Chen, and K. Shimada, “Vision-aided uav navigation and dynamic obstacle avoidance using gradient-based b- spline trajectory optimization,” in 2023 IEEE International Conference on Robotics and Automation (ICRA) , pp. 1214–1220. IEEE, 2023

work page 2023

Show all 22 references

[9]

Robust vision-based obstacle avoidance for micro aerial vehicles in dynamic environments,

J. Lin, H. Zhu, and J. Alonso-Mora, “Robust vision-based obstacle avoidance for micro aerial vehicles in dynamic environments,” in 2020 IEEE International Conference on Robotics and Automation (ICRA) , pp. 2682–2688. IEEE, 2020

2020
[10]

Dynamic obstacle avoid- ance for quadrotors with event cameras,

D. Falanga, K. Kleber, and D. Scaramuzza, “Dynamic obstacle avoid- ance for quadrotors with event cameras,”Science Robotics, vol. 5, no. 40, p. eaaz9712, 2020

2020
[11]

Avoiding dynamic small obstacles with onboard sensing and computation on aerial robots,

F. Kong, W. Xu, Y . Cai, and F. Zhang, “Avoiding dynamic small obstacles with onboard sensing and computation on aerial robots,” IEEE Robotics and Automation Letters , vol. 6, no. 4, pp. 7869–7876, 2021

2021
[12]

Fapp: Fast and adaptive perception and planning for uavs in dynamic cluttered environments,

M. Lu, X. Fan, H. Chen, and P. Lu, “Fapp: Fast and adaptive perception and planning for uavs in dynamic cluttered environments,”arXiv preprint arXiv:2312.08743, 2023

2023 arXiv
[13]

Moving event detection from lidar point streams,

H. Wu, Y . Li, W. Xu, F. Kong, and F. Zhang, “Moving event detection from lidar point streams,” nature communications, vol. 15, no. 1, p. 345, 2024

2024
[14]

Integrated planning and control for quadrotor navigation in presence of suddenly appearing objects and disturbances,

W. Liu, Y . Ren, and F. Zhang, “Integrated planning and control for quadrotor navigation in presence of suddenly appearing objects and disturbances,” IEEE Robotics and Automation Letters , 2023

2023
[15]

Fast-dynamic-vision: Detection and tracking dynamic objects with event and depth sensing,

B. He, H. Li, S. Wu, D. Wang, Z. Zhang, Q. Dong, C. Xu, and F. Gao, “Fast-dynamic-vision: Detection and tracking dynamic objects with event and depth sensing,” in 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pp. 3071–3078. IEEE, 2021

2021
[16]

Dpmpc-planner: A real- time uav trajectory planning framework for complex static environments with dynamic obstacles,

Z. Xu, D. Deng, Y . Dong, and K. Shimada, “Dpmpc-planner: A real- time uav trajectory planning framework for complex static environments with dynamic obstacles,” in 2022 International Conference on Robotics and Automation (ICRA) , pp. 250–256. IEEE, 2022

2022
[17]

A density-based algorithm for discovering clusters in large spatial databases with noise,

M. Ester, H.-P. Kriegel, J. Sander, X. Xu et al. , “A density-based algorithm for discovering clusters in large spatial databases with noise,” in kdd, vol. 96, no. 34, pp. 226–231, 1996

1996
[18]

The hungarian method for the assignment problem,

H. W. Kuhn, “The hungarian method for the assignment problem,” Naval research logistics quarterly , vol. 2, no. 1-2, pp. 83–97, 1955. IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS

1955
[19]

Rog-map: An efficient robocentric occupancy grid map for large-scene and high-resolution lidar-based motion planning,

Y . Ren, Y . Cai, F. Zhu, S. Liang, and F. Zhang, “Rog-map: An efficient robocentric occupancy grid map for large-scene and high-resolution lidar-based motion planning,” arXiv preprint arXiv:2302.14819 , 2023

2023 arXiv
[20]

A formal basis for the heuristic determination of minimum cost paths,

P. E. Hart, N. J. Nilsson, and B. Raphael, “A formal basis for the heuristic determination of minimum cost paths,” IEEE transactions on Systems Science and Cybernetics , vol. 4, no. 2, pp. 100–107, 1968. Huajie Wu (Graduate Student Member, IEEE) received the B.Eng. degree in A...

1968
[21]

Osqp: An operator splitting solver for quadratic programs,

B. Stellato, G. Banjac, P. Goulart, A. Bemporad, and S. Boyd, “Osqp: An operator splitting solver for quadratic programs,” Mathematical Programming Computation, vol. 12, no. 4, pp. 637–672, 2020

2020
[22]

Fast-lio2: Fast direct lidar- inertial odometry,

W. Xu, Y . Cai, D. He, J. Lin, and F. Zhang, “Fast-lio2: Fast direct lidar- inertial odometry,” IEEE Transactions on Robotics , vol. 38, no. 4, pp. 2053–2073, 2022. Hairuo Wei received the B.Eng. degree in au- tomation, in 2023, from Harbin Institute of Tech- nology, Shenzhen,...

2022

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.

[1] [1]

Adoption of unmanned aerial vehicle (uav) imagery in agricultural management: A systematic literature review,

M. A. Istiak, M. M. Syeed, M. S. Hossain, M. F. Uddin, M. Hasan, R. H. Khan, and N. S. Azad, “Adoption of unmanned aerial vehicle (uav) imagery in agricultural management: A systematic literature review,” Ecological Informatics , p. 102305, 2023

work page 2023

[2] [2]

Airborne! uav delivery of blood products and medical logistics for combat zones,

D. T. Lammers, J. M. Williams, J. R. Conner, E. Baird, O. Rokayak, J. M. McClellan, J. R. Bingham, R. Betzold, and M. J. Eckert, “Airborne! uav delivery of blood products and medical logistics for combat zones,” Transfusion, vol. 63, pp. S96–S104, 2023

work page 2023

[3] [3]

LiDAR-based Quadrotor for Slope Inspection in Dense Vegetation

W. Liu, Y . Ren, R. Guo, V . W. Kong, A. S. Hung, F. Zhu, Y . Cai, Y . Zou, and F. Zhang, “Lidar-based quadrotor for slope inspection in dense vegetation,” arXiv preprint arXiv:2409.13985 , 2024

work page Pith review arXiv 2024

[4] [4]

Unmanned aerial vehicles for search and rescue: A survey,

M. Lyu, Y . Zhao, C. Huang, and H. Huang, “Unmanned aerial vehicles for search and rescue: A survey,” Remote Sensing , vol. 15, no. 13, p. 3266, 2023

2023

[5] [5]

Panther: Perception-aware trajectory planner in dynamic environments,

J. Tordesillas and J. P. How, “Panther: Perception-aware trajectory planner in dynamic environments,” IEEE Access , vol. 10, pp. 22 662– 22 677, 2022

2022

[6] [6]

Perception and avoidance of multiple small fast moving objects for quadrotors with only low-cost rgbd camera,

M. Lu, H. Chen, and P. Lu, “Perception and avoidance of multiple small fast moving objects for quadrotors with only low-cost rgbd camera,” IEEE Robotics and Automation Letters , vol. 7, no. 4, pp. 11 657–11 664, 2022

work page 2022

[7] [7]

Flying in dynamic scenes with multitarget velocimetry and perception-enhanced planning,

H. Chen, C.-Y . Wen, F. Gao, and P. Lu, “Flying in dynamic scenes with multitarget velocimetry and perception-enhanced planning,”IEEE/ASME Transactions on Mechatronics , 2023

work page 2023

[8] [8]

Vision-aided uav navigation and dynamic obstacle avoidance using gradient-based b- spline trajectory optimization,

Z. Xu, Y . Xiu, X. Zhan, B. Chen, and K. Shimada, “Vision-aided uav navigation and dynamic obstacle avoidance using gradient-based b- spline trajectory optimization,” in 2023 IEEE International Conference on Robotics and Automation (ICRA) , pp. 1214–1220. IEEE, 2023

work page 2023