Pith. sign in

REVIEW 4 major objections 5 minor 29 references

Quadrotor Morpho-Transition: Learning vs Model-Based Control Strategies

T0 review · 4 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read A reinforcement-learning controller lands a morphing quadrotor at a 65-degree tilt, beating an MPC baseline that reaches only 60 degrees.

desk verdict First hardware demo of end-to-end RL for mid-air morpho-transition, with a fair but thin comparison against MPC; the core result stands, but the paper's own admission about latency/motor-dynamics mismatch undermines its strongest sim-to-real claim. read the letter →

arxiv 2506.14039 v1 pith:YPJA5LOM submitted 2025-06-16 cs.RO cs.SYeess.SY

Ioannis Mandralis , Richard M. Murray , Morteza Gharib This is my paper · ORCID
classification cs.ROcs.SYeess.SY
keywords quadrotormorpho-transitionreinforcementlearningsim-to-realtransfermodelpredictivecontroldomainrandomizationactuatorfailurerecoveryaerial-groundtransitiondisturbancerejection
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

The paper aims to establish that an end-to-end reinforcement-learning controller can perform quadrotor morpho-transition—transforming in mid-air from a flying quadrotor into a wheeled rover and landing on those wheels—and that it can outperform a model-predictive-control baseline on real hardware. The central claim is that the RL policy transfers from simulation to the physical ATMO robot only when the simulator models first-order motor lag with a 0.15 s time constant and a lumped 20 ms observation delay. With those in place, the policy lands at a 65-degree tilt with a 0.5 m/s impact velocity, while the MPC baseline reaches 60 degrees at 1.0 m/s. The paper also claims the learned policy rejects small and moderate pushes better than MPC and continues to recover from substantial pushes under partial rotor loss, while MPC recovers only from the very largest pushes. If these claims hold, mid-air transformation control no longer needs hand-engineered contact planning or manually tuned attitude loops.

What carries the argument

The load-bearing mechanism is the simulator's hardware-relevant loop closure: the robot is modeled as a rigid body whose tilt angle obeys a pure integrator $\dot{\phi} = \dot{\phi}_{\max} u_{\mathrm{body}}$, whose four rotors obey a first-order motor filter $\dot{\Omega} = T_m^{-1}(u_{\mathrm{aero}} - \Omega)$ with nominal $T_m = 0.15$ s, and whose policy observations are delayed by a single 20 ms step. Around this core, the authors randomize the motor time constant over $[0.10, 0.20]$ s, the thrust and moment coefficients by ±20%, the tilt speed, and apply one random push per episode; the reward is shaped to land on the wheels near the goal while penalizing contact impulse and thrust during ground contact, and the critic receives privileged information about disturbance forces, push timing, and contact impulse. A full rotation-matrix observation, rather than a quaternion, is also required for transfer. The decisive claim is that removing either the motor filter or the observation delay breaks sim-to-real transfer.

What would settle it

Measure the actual step response of an ATMO rotor from commanded RPM to measured thrust, and the actual end-to-end observation delay; if the best-fit motor time constant falls outside the trained randomization range of 0.10–0.20 s, or the latency is not close to a single 20 ms lumped delay, the paper's explanation for successful transfer is falsified. A second direct test: run the trained policy on hardware with an extra 40 ms of artificial observation delay; if it still lands at the reported tilt and impact velocity, then observation-delay modeling is not the load-bearing element claimed.

Watch

Extended reading notes

Core claim

The central claim is that a single neural-network policy, trained from raw observations straight to motor-speed and tilt-velocity commands, can carry ATMO through a morpho-transition, and that the deciding factor for hardware transfer is not the algorithm but simulation fidelity in three specific places: the first-order response of the rotors, a one-step 20 ms delay on observations, and randomized motor, thrust, tilt, and disturbance parameters. The trained policy lands at 65 degrees, past the 60-degree tilt at which the thrusters saturate, with a lower impact velocity than the MPC baseline, and in simulation it absorbs push disturbances up to roughly five to six times an individual rotor's thrust—beyond what it was trained on—before degrading. Under a partial rotor failure that scales the four rotor coefficients by [0.8, 0.9, 0.85, 1.1], the RL policy still recovers from substantial pushes, whereas the MPC baseline fails nearly all such tests. The trade-off visible in the hardware data is that the RL landing has larger roll oscillations, which the authors attribute to imperfect knowledge of latency and the motor time constant.

Load-bearing premise

The load-bearing premise is that the simplified simulator—stiff tilt integrator, first-order motors with a 0.15 s time constant, one 20 ms observation delay, and no higher-order aerodynamics—is faithful enough that a policy trained on it behaves the same on the real ATMO; if any of those approximations misrepresent the hardware, the claimed transfer fails or the landing degrades.

Editorial extensions

If this is right

  • Morpho-transition can be solved as one end-to-end learned task: no separate flight and ground phases, no contact-phase switching, and no hand-tuned attitude loops are needed for a successful wheel landing.
  • The same training recipe—include motor dynamics and a one-step observation delay, then randomize dynamics—should extend to other tilting-rotor or foldable quadrotors that change thrust orientation mid-air.
  • Because the learned policy operates beyond the 60-degree static saturation tilt, morpho-transition can reach deeper drive-like configurations, expanding the usable landing envelope for confined spaces.
  • The policy's recovery from partial rotor loss without explicit fault detection demonstrates a graceful-degradation mode for learned flight controllers, within a bounded disturbance region.
  • The measured split—good small-push recovery for RL, good large-push recovery for MPC—maps a concrete operating envelope for each controller rather than declaring a single winner.

Reading between the lines

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

  • If the roll oscillation comes from residual latency mismatch, moving policy inference onto the flight controller and cutting the communication hop should reduce it without retraining; the paper's own diagnosis makes this a testable prediction.
  • The complementary failure regions point to a natural hybrid: let the learned policy handle the precise final landing phase, and let a model-based planner take over when the estimated disturbance exceeds a threshold.
  • The same simulation design—first-order actuation, a fixed observation delay, and randomization, rather than more detailed physics—may transfer to other contact-rich aerial maneuvers such as perching or hopping.
  • An immediate extension would be to train under correlated rotor failures rather than only uniform coefficient randomization, to test whether the observed actuator-failure recovery is a general robustness property or an artifact of the training distribution.
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.

Referee Report

4 major / 5 minor

Summary. The paper studies the morpho-transition maneuver for the ATMO quadrotor, comparing an end-to-end RL policy trained with PPO and domain randomization in Isaac Lab against the model-based MPC controller from the authors' prior work. The central claims are that the RL policy transfers to hardware, achieves a final tilt angle of 65 degrees with an impact velocity of 0.5 m/s (versus 60 degrees and 1.0 m/s for MPC), exhibits better disturbance rejection than MPC for small-to-moderate pushes in simulation, and recovers from partial actuator failures without explicit fault knowledge. The authors additionally claim that the RL policy transfers to hardware only because the simulator includes motor dynamics and a 20 ms observation delay. The paper contributes an open-source simulator extension, a benchmark protocol, and a hardware demonstration.

Significance. If the claims hold, this is a useful contribution: it is one of the first demonstrations of end-to-end RL for mid-air transformation, it provides a principled comparison with an MPC baseline, and the code release and simulator extension are valuable for the community. The strength of the paper is its explicit attention to the ingredients needed for sim-to-real transfer, namely motor dynamics, observation delays, and domain randomization. The central result, however, rests on a single hardware demonstration per controller and on simulation-only benchmarks whose predictive value for the real robot is not established, so the quantitative claims should be treated with caution until the evidentiary basis is strengthened.

major comments (4)
  1. [Sec. IV, Fig. 5] The hardware comparison is based on one representative run per controller, with no trial counts, success criteria, or error statistics. The central quantitative claims (RL final tilt 65 deg vs. MPC 60 deg; impact velocity 0.5 m/s vs. 1.0 m/s) are therefore not statistically supported. Please report multiple trials per controller, define a success/failure criterion, and report mean and variance, or explicitly characterize the hardware section as a single demonstration.
  2. [Sec. IV, final paragraph] The paper states that the RL controller 'exhibits larger oscillations in the roll angle' and attributes this to 'inaccurate estimation of the system latency as well as the motor dynamics time constant.' This is a self-reported limitation on exactly the two simulator ingredients that Sec. III.D and the abstract claim are necessary for transfer, namely the 20 ms observation delay and Tm = 0.15 s. The observation weakens the causal claim: the policy may transfer because of robustness from domain randomization rather than because those parameters are accurately modeled. Please provide hardware measurements of the actual delay and motor time constant, or re-frame the claim as 'the policy transfers when trained with these modeled effects' rather than 'only transfers because these effects are accurately captured.'
  3. [Secs. V and VI] All quantitative comparisons of disturbance recovery and actuator failure are performed in the same simplified simulator used for training, with a single rollout per condition and no variance reporting; the paper itself defers hardware characterization to future work (Sec. VII). The statement in Sec. VII that 'the RL method performs better at small disturbances' should therefore be labeled as a simulation-only result. Please report how many random seeds and initial conditions were used per push condition, and add error bars or confidence intervals.
  4. [Sec. III.A] The closed-loop morphing linkage is modeled as a pure tilt-velocity integrator with a stiff PD controller (kp = 1e15, kd = 1e5), while the paper acknowledges that the real mechanism is difficult to simulate. Because the tilt dynamics and ground-contact behavior are central to the morpho-transition maneuver, this unvalidated simplification is a threat to the sim-to-real explanation. Please provide sensitivity tests of the policy to variations in the tilt-mechanism model or justify the simplification with hardware measurements.
minor comments (5)
  1. [Eq. (11)] The symbol phi_d is used in Eq. (11) but is not defined before its use in Sec. III.A; please define the desired tilt angle and its relationship to ubody.
  2. [Sec. III.E] The text says the reward coefficients a1...a9 are given in the appendix, but Eq. (14) uses a0; the appendix correctly lists a0...a9. Please align the notation.
  3. [Fig. 6] The figure reports an 'interpolated heat map' but no color scale or axis labels are visible; please add a colorbar and define the units for impact velocity and distance to goal.
  4. [Sec. III.B] The statement that 'training without randomizing these parameters results in neither sim-to-sim nor sim-to-real transfer' is a strong empirical claim with no supporting figure or table; please substantiate it with data or soften the wording.
  5. [Abstract and Sec. IV] The abstract says the RL controller 'only transfers to hardware if motor dynamics and observation delays are taken into account,' while Sec. IV reports that the hardware roll oscillations are attributed to inaccurate estimation of exactly those quantities; please reconcile these statements in the final version.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: empirical transfer claims with a fair self-cited baseline; modeling limitations are validation concerns, not circular reductions.

full rationale

The paper's central claims are empirical rather than derivational: an end-to-end RL policy transfers to ATMO hardware, and its disturbance-rejection behavior is compared with an MPC baseline. The claim that motor dynamics and observation delays are necessary for transfer is supported by ablations and hardware experiments, not by construction. The MPC baseline is the authors' prior work [1], but it is used as a fair comparator in the same simulator and on the same hardware, and no load-bearing conclusion is reduced to that self-citation. The simulator model follows [1] and explicitly simplifies the closed-loop morphing mechanism to a pure integrator with stiff PD tracking, first-order motor dynamics with Tm = 0.15 s, and a 20 ms observation delay; these are transparent modeling assumptions, and the paper's own Sec. IV admits they may be inaccurate ('We believe this is due to inaccurate estimation of the system latency as well as the motor dynamics time constant'). That admission weakens the model-fidelity explanation for transfer, but it is a correctness or validation limitation, not circularity. The reward function does shape the RL policy to descend at 0.5 m/s and to seek the goal, so the reported 0.5 m/s impact velocity partly reflects reward design; however, the paper is transparent about this reward, and the main transfer and comparison claims are externally evidenced by hardware experiments. No equation or definition reduces to its own output, and no fitted parameter is renamed as a prediction. The score of 1 reflects only the minor presence of the authors' own prior work as baseline and model source; it is not load-bearing circularity.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claims rest on hand-tuned reward coefficients, hand-chosen domain randomization ranges, and the assumption that the simplified simulator, including the stiff morphing mechanism model and a single 20ms delay, is an accurate surrogate for the real robot. No new physical entities are introduced; ATMO and its MPC are from prior work [1].

free parameters (3)
  • Reward coefficients a0..a9 = a0=-0.10Ts, a1=-0.30Ts, a2=-0.10Ts, a3=-0.07Ts, a4=-0.13Ts, a5=-1.0, a6=0.40, a7=0.30Ts, a8=0.40Ts, a9=0.30Ts
    Hand-tuned to encode landing behavior; the paper does not report an automated tuning procedure, and the learned behavior depends on them.
  • Domain randomization ranges = f*=0.20cT, tau*=0.20cT*cM, Tm~U(0.10,0.20), cT/cM ±20%, phi_dot_max ±20%
    Chosen by hand; the paper states training without these ranges yields neither sim-to-sim nor sim-to-real transfer, so they are load-bearing.
  • Motor time constant and observation delay nominal values = Tm=0.15s, delay=20ms
    Estimated for the hardware; the paper later attributes RL roll oscillations partly to inaccurate estimation of these values, so they function as fit parameters for transfer.
assumptions (4)
  • domain assumption The morphing mechanism can be modeled as a pure integrator for tilt angle with a stiff PD tracker (Sec III.A), and the closed-loop kinematic linkage is effectively rigid.
    The actual kinematic chain is 'difficult to simulate' (Sec III.A) and is replaced by a high-gain PD; if the real mechanism has compliance or unmodeled dynamics, the policy may not transfer.
  • domain assumption The simplified dynamics model (Eq 8-11) with linear thrust/moment, first-order motor dynamics, and a single 20ms observation delay is sufficient to capture real ATMO behavior for transfer.
    Sec III.D states this was found necessary, and Sec IV notes residual roll oscillations from inaccurate latency/motor time constant estimates.
  • domain assumption ATMO's symmetry allows restricting disturbance tests to the first quadrant of the body xy plane (Sec V).
    Assumes the robot and the policy are symmetric about the sagittal plane for this maneuver; not validated on hardware.
  • domain assumption Higher-order aerodynamic effects (proximity, flapping, rotor couplings) can be ignored (Sec II.B).
    Standard for this line of work; the paper acknowledges it.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Quadrotor Morpho-Transition: Learning vs Model-Based Control Strategies." pith.science (2026). https://pith.science/paper/YPJA5LOM

@misc{pith2026250614039,
  author       = {Pith},
  title        = {Pith review of: Quadrotor Morpho-Transition: Learning vs Model-Based Control Strategies},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YPJA5LOM}},
  note         = {Machine review of arXiv:2506.14039}
}
read the original abstract

Quadrotor Morpho-Transition, or the act of transitioning from air to ground through mid-air transformation, involves complex aerodynamic interactions and a need to operate near actuator saturation, complicating controller design. In recent work, morpho-transition has been studied from a model-based control perspective, but these approaches remain limited due to unmodeled dynamics and the requirement for planning through contacts. Here, we train an end-to-end Reinforcement Learning (RL) controller to learn a morpho-transition policy and demonstrate successful transfer to hardware. We find that the RL control policy achieves agile landing, but only transfers to hardware if motor dynamics and observation delays are taken into account. On the other hand, a baseline MPC controller transfers out-of-the-box without knowledge of the actuator dynamics and delays, at the cost of reduced recovery from disturbances in the event of unknown actuator failures. Our work opens the way for more robust control of agile in-flight quadrotor maneuvers that require mid-air transformation.

Figures

Figures reproduced from arXiv: 2506.14039 by the authors.

Figure 1
Figure 1. Snapshots of ATMO performing the Morpho-Transition maneuver [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. A photo of ATMO, the Aerially Transforming Morphobot, in ground [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. (a) Front-view of ATMO in flight mode with thrust [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: A depiction of the closed loop kinematic linkage enabling symmetric [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: Morpho-Transition maneuver performed on the hardware using the two controllers. The performance of the MPC controller is shown on the left, [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]
Figure 6
Figure 6. Figure 6: Comparison between RL and MPC methods. The impact velocity and final distance to goal was recorded for 64 different push forces for each [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]
Figure 7
Figure 7. Figure 7: Recovery characteristics under partial actuator failure. The [PITH_FULL_IMAGE:figures/full_fig_p006_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

29 extracted references · 15 canonical work pages

  1. [1]

    ATMO: an aerially transforming morphobot for dynamic ground-aerial transition,

    I. Mandralis, R. Nemovi, A. Ramezani, R. M. Murray, and M. Gharib, “ATMO: an aerially transforming morphobot for dynamic ground-aerial transition,” Communications Engineering , vol. 4, no. 1, Apr. 2025. [Online]. Available: http://dx.doi.org/10.1038/ s44172-025-00413-6

    work page 2025

  2. [2]

    The foldable drone: A morphing quadrotor that can squeeze and fly,

    D. Falanga, K. Kleber, S. Mintchev, D. Floreano, and D. Scaramuzza, “The foldable drone: A morphing quadrotor that can squeeze and fly,” IEEE Robotics and Automation Letters , vol. 4, no. 2, pp. 209–216, 2019

    2019

  3. [3]

    X-morf: A crash-separable quadrotor that morfs its x-geometry in flight,

    A. Desbiez, F. Expert, M. Boyron, J. Diperi, S. Viollet, and F. Ruffier, “X-morf: A crash-separable quadrotor that morfs its x-geometry in flight,” in 2017 Workshop on Research, Education and Development of Unmanned Aerial Systems (RED-UAS) , 2017, pp. 222–227

    2017

  4. [4]

    Design and control of a passively morphing quadcopter,

    N. Bucki and M. W. Mueller, “Design and control of a passively morphing quadcopter,” in 2019 International Conference on Robotics and Automation (ICRA) , 2019, pp. 9116–9122

    work page 2019

  5. [5]

    Design and control of a midair-reconfigurable quadcopter using unactuated hinges,

    N. Bucki, J. Tang, and M. W. Mueller, “Design and control of a midair-reconfigurable quadcopter using unactuated hinges,” 2021. [Online]. Available: https://arxiv.org/abs/2103.16632

    arXiv 2021

  6. [6]

    Minimum time trajectory generation for bounding flight: Combining posture control and thrust vectoring,

    I. Mandralis, E. Sihite, A. Ramezani, and M. Gharib, “Minimum time trajectory generation for bounding flight: Combining posture control and thrust vectoring,” in 2023 European Control Conference (ECC) , 2023, pp. 1–7

    work page 2023

  7. [7]

    Control of a quadrotor with reinforcement learning,

    J. Hwangbo, I. Sa, R. Siegwart, and M. Hutter, “Control of a quadrotor with reinforcement learning,” IEEE Robotics and Automation Letters , vol. 2, no. 4, pp. 2096–2103, 2017

    2017

  8. [8]

    Champion-level drone racing using deep reinforcement learning,

    E. Kaufmann, L. Bauersfeld, A. Loquercio, M. M ¨uller, V . Koltun, and D. Scaramuzza, “Champion-level drone racing using deep reinforcement learning,” Nature, vol. 620, no. 7976, p. 982–987, Aug. 2023. [Online]. Available: http://dx.doi.org/10.1038/s41586-023-06419-4

    doi:10.1038/s41586-023-06419-4 2023

Show all 29 references
  1. [9]

    Sim-to-(multi)-real: Transfer of low-level robust control policies to multiple quadrotors,

    A. Molchanov, T. Chen, W. H ¨onig, J. A. Preiss, N. Ayanian, and G. S. Sukhatme, “Sim-to-(multi)-real: Transfer of low-level robust control policies to multiple quadrotors,” 2019. [Online]. Available: https://arxiv.org/abs/1903.04628

    2019 arXiv

  2. [10]

    A benchmark comparison of learned control policies for agile quadrotor flight,

    E. Kaufmann, L. Bauersfeld, and D. Scaramuzza, “A benchmark comparison of learned control policies for agile quadrotor flight,” in 2022 International Conference on Robotics and Automation (ICRA) , 2022, pp. 10 504–10 510

    2022

  3. [11]

    Learning to fly in seconds,

    J. Eschmann, D. Albani, and G. Loianno, “Learning to fly in seconds,” 2023. [Online]. Available: https://arxiv.org/abs/2311.13081

    2023 arXiv

  4. [12]

    A learning-based quadcopter controller with extreme adaptation,

    D. Zhang, A. Loquercio, J. Tang, T.-H. Wang, J. Malik, and M. W. Mueller, “A learning-based quadcopter controller with extreme adaptation,” 2024. [Online]. Available: https://arxiv.org/abs/2409. 12949

    2024

  5. [13]

    Autonomous quadrotor landing using deep reinforcement learning,

    R. Polvara, M. Patacchiola, S. Sharma, J. Wan, A. Manning, R. Sutton, and A. Cangelosi, “Autonomous quadrotor landing using deep reinforcement learning,” 2017. [Online]. Available: https://arxiv.org/abs/1709.03339

    2017 arXiv

  6. [14]

    Inclined quadrotor landing using deep reinforcement learning,

    J. E. Kooi and R. Babu ˇska, “Inclined quadrotor landing using deep reinforcement learning,” in 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , 2021, pp. 2361–2368

    2021

  7. [15]

    Lander.ai: Adaptive landing behavior agent for expertise in 3d dynamic platform landings,

    R. Peter, L. Ratnabala, D. Aschu, A. Fedoseev, and D. Tsetserukou, “Lander.ai: Adaptive landing behavior agent for expertise in 3d dynamic platform landings,” 2024. [Online]. Available: https: //arxiv.org/abs/2403.06572

    2024 arXiv

  8. [16]

    Robot operating system 2: Design, architecture, and uses in the wild,

    S. Macenski, T. Foote, B. Gerkey, C. Lalancette, and W. Woodall, “Robot operating system 2: Design, architecture, and uses in the wild,” Science Robotics , vol. 7, no. 66, May 2022. [Online]. Available: http://dx.doi.org/10.1126/scirobotics.abm6074

    2022 doi

  9. [17]

    proNEu: Derivation of ana- lytical kinematics and dynamics,

    M. Hutter, C. Gehring, and R. Siegwart, “proNEu: Derivation of ana- lytical kinematics and dynamics,” Autonomous Systems Lab, ETHZ, Tech. Rep., 2011

    2011

  10. [18]

    The role of propeller aerodynamics in the model of a quadrotor UA V,

    P.-J. Bristeau, P. Martin, E. Salaun, and N. Petit, “The role of propeller aerodynamics in the model of a quadrotor UA V,” in 2009 European Control Conference (ECC) . IEEE, Aug. 2009. [Online]. Available: http://dx.doi.org/10.23919/ECC.2009.7074482

    2009

  11. [19]

    A multiple shooting algorithm for direct solution of optimal control problems *,

    H. Bock and K. Plitt, “A multiple shooting algorithm for direct solution of optimal control problems *,” IFAC Proceedings Volumes, vol. 17, no. 2, p. 1603–1608, Jul. 1984. [Online]. Available: http://dx.doi.org/10.1016/S1474-6670(17)61205-9

    1984 doi

  12. [20]

    acados - a modular open-source framework for fast embedded optimal control,

    R. Verschueren, G. Frison, D. Kouzoupis, J. Frey, N. van Duijkeren, A. Zanelli, B. Novoselnik, T. Albin, R. Quirynen, and M. Diehl, “acados - a modular open-source framework for fast embedded optimal control,” Mathematical Programming Computation, vol. 14, pp. 147 – 183, 2019....

    2019

  13. [21]

    CasADi: a software framework for nonlinear optimization and optimal control,

    J. A. E. Andersson, J. Gillis, G. Horn, J. B. Rawlings, and M. Diehl, “CasADi: a software framework for nonlinear optimization and optimal control,” Mathematical Programming Computation, vol. 11, no. 1, p. 1–36, Jul. 2018. [Online]. Available: http://dx.doi.org/10.1007/s12532-...

    2018 doi

  14. [22]

    Isaac gym: High performance gpu-based physics simulation for robot learning,

    V . Makoviychuk, L. Wawrzyniak, Y . Guo, M. Lu, K. Storey, M. Macklin, D. Hoeller, N. Rudin, A. Allshire, A. Handa, and G. State, “Isaac gym: High performance gpu-based physics simulation for robot learning,” 2021. [Online]. Available: https://arxiv.org/abs/2108.10470

    2021 arXiv

  15. [23]

    Orbit: A unified simulation framework for interactive robot learning environments,

    M. Mittal, C. Yu, Q. Yu, J. Liu, N. Rudin, D. Hoeller, J. L. Yuan, R. Singh, Y . Guo, H. Mazhar, A. Mandlekar, B. Babich, G. State, M. Hutter, and A. Garg, “Orbit: A unified simulation framework for interactive robot learning environments,” IEEE Robotics and Automa- tion Lette...

    2023

  16. [24]

    Cat-like jumping and landing of legged robots in low gravity using deep reinforcement learning,

    N. Rudin, H. Kolvenbach, V . Tsounis, and M. Hutter, “Cat-like jumping and landing of legged robots in low gravity using deep reinforcement learning,” IEEE Transactions on Robotics, vol. 38, no. 1, pp. 317–328, 2022

    2022

  17. [25]

    In-flight attitude control of a quadruped using deep reinforcement learning,

    T. El-Agroudi, F. G. Maurer, J. A. Olsen, and K. Alexis, “In-flight attitude control of a quadruped using deep reinforcement learning,” in 8th Annual Conference on Robot Learning , 2024. [Online]. Available: https://openreview.net/forum?id=67tTQeO4HQ

    2024

  18. [26]

    Asymmetric actor critic for image-based robot learning,

    L. Pinto, M. Andrychowicz, P. Welinder, W. Zaremba, and P. Abbeel, “Asymmetric actor critic for image-based robot learning,” 2017. [Online]. Available: https://arxiv.org/abs/1710.06542

    2017 arXiv

  19. [27]

    Proximal policy optimization algorithms,

    J. Schulman, F. Wolski, P. Dhariwal, A. Radford, and O. Klimov, “Proximal policy optimization algorithms,” 2017. [Online]. Available: https://arxiv.org/abs/1707.06347

    2017 arXiv

  20. [28]

    PX4: A node-based multithreaded open source robotics framework for deeply embedded platforms,

    L. Meier, D. Honegger, and M. Pollefeys, “PX4: A node-based multithreaded open source robotics framework for deeply embedded platforms,” in 2015 IEEE International Conference on Robotics and Automation (ICRA), 2015, pp. 6235–6240

    2015

  21. [29]

    Nonlinear MPC for quadrotor fault-tolerant control,

    F. Nan, S. Sun, P. Foehn, and D. Scaramuzza, “Nonlinear MPC for quadrotor fault-tolerant control,” IEEE Robotics and Automation Letters, vol. 7, no. 2, pp. 5047–5054, 2022. VIII. A PPENDIX A. Reward function coefficients Coefficient Value Description a0 −0.10Ts Velocity Penalt...

    2022

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.