Chapter Four · failure evidence
What Feedback & Closed-Loop Control got wrong, from 86 dissertations
These records demonstrate the operational trade-offs between open-loop execution and closed-loop feedback across robotics, chemical processes, and mechatronic systems. While open-loop methods frequently suffer from drift and disturbance accumulation, closed-loop feedback systems can fail due to latency, excessive gain tuning, sensor noise, or unmodeled plant interactions. These records come from PhD theses at 28 institutions, 2021 to 2026. Each links to its thesis. They were extracted by language models reading the full text, so treat each as a lead to read, not a verdict.
Open-loop control fails due to dynamic disturbances, drift, and lack of real-time state correction
Operating without closed-loop feedback leads to accumulated trajectory drift, unbounded covariance growth, and failure to track references under environmental uncertainty or contact variability. In multiple evaluations, open-loop policies were abandoned or lost to closed-loop controllers because they could not regulate against unmodeled dynamics and disturbances.
Tried and failed
open-loop optimal control during initialization phase applied to system startup trajectory tracking. Outcome: worse than baseline. Reason: lack of feedback during startup failed to outperform standard closed-loop trajectory tracking
Automated Optimization and Control of Modular Chemical Systems · MIT
Tried and failed
open-loop model predictive control using constraint-based models applied to bioreactor nutrient feed rate optimization. Reason: unmodeled environmental dynamics and residual metabolism caused nutrient accumulation and growth deviations without feedback
Genome-scale model-based approach to optimize nutrient supply for chlorella vulgaris · JScholarship
Tried and failed
open-loop transfer of simulation-optimized trajectories to hardware applied to locomotion controller for physical robots. Outcome: did not generalise. Reason: sim-to-real gap and unmodeled contact dynamics caused unpredictable execution without feedback
Robot Graph Grammars: Towards Custom Robots for Every Task · MIT
Tried and failed
open-loop actuation for manipulating flexible materials applied to automated loop lifting in robotic crocheting. Outcome: unstable. Reason: loose, deformable fabric introduced spatial variability, causing mechanical actuators to miss loops or lift inadequately
Croche-Matic – building a robot for crocheting 3D spherical form · Harvard
Lost to a baseline
The open-loop stochastic self-triggered MPC consumed significantly more resource and stayed farther from reference boundaries than the closed-loop feedback formulation.
Data-driven Methods for Control: from Linear to Lifting · EPFL
Lost to a baseline
Open-loop trajectory tracking of hydraulic actuator position resulted in substantially higher tracking error (16.65° RMSE for sinusoidal references) compared to closed-loop PI control (3.90° RMSE).
Considered and rejected
Considered and rejected: Rejected open-loop / feed-forward GPS/INS integration configurations because reference and actual trajectories diverge, deteriorating linearity and process model assumptions over time; adopted feedback/closed-loop integration instead.
Vision based Real-Time Navigation with Unknown and Uncooperative Space Target · Carleton University Institutional Repository
Considered and rejected
Considered and rejected: Rejected open-loop charge execution at t=0 in favor of closed-loop MPC-style periodic recomputation to handle forecast disturbances, price changes, and parameter errors.
Smart Charging of Electric Vehicles: Algorithms, Ramifications, and Hardware Development · Georgia Tech
Considered and rejected
Considered and rejected: Rejected open-loop AOM RF frequency/amplitude control for lattice ramping due to drift and complexity, retaining closed-loop PID control
Considered and rejected
Considered and rejected: Open-loop RF frequency/amplitude control of the AOM for deep lattice ramping due to drift and complexity, retaining closed-loop PID control
Considered and rejected
Considered and rejected: Open-loop proportional distance control alone: rejected because steady-state following error grew with target speed, risking tag loss; replaced with PI distance control.
USE OF ARTIFICIAL FIDUCIAL MARKERS FOR USV SWARM COORDINATION · Calhoun
Considered and rejected
Considered and rejected: Rejected open-loop state covariance propagation for rendezvous chance constraints in favor of a feedback-augmented deadbeat impulsive controller to suppress unbounded covariance growth.
Fast and Resilient Optimization-based Control with Spacecraft Applications · ResearchWorks
Tried and failed
open-loop pulse programming of analog states applied to resistive switching memory devices. Outcome: unstable. Reason: stochastic switching prevented reproducible target resistance states and reliable retention without closed-loop verification
High-resolution peripheral circuits for multi-state memristors and their related applications · Imperial
Tried and failed
open-loop rate control transitions applied to vehicle trajectory planning. Outcome: unstable. Reason: lack of feedback during transitions accumulated trajectory drift and positional error
Adapting Random Tree Search to Fixed Wing Aerial Vehicles with Closed-Loop Prediction and Hybrid Control · Georgia Tech
Tried and failed
open-loop fixed timing switching control applied to resonant soft-switching power converters. Outcome: unstable. Reason: fixed delays failed during transients, causing hard-switching transitions and destructive overvoltage stresses
Multiport Power Electronics based on Four-Quadrant Soft-Switching Current-Source Converters · Georgia Tech
Tried and failed
open-loop robust control with uncertainty propagation applied to dynamic ambiguity set control. Outcome: unstable. Reason: Zero feedback gain fails to restrict the unbounded growth of dynamic uncertainty sets over time.
Data-driven distributionally robust model predictive control · Imperial
Lost to a baseline
Open-loop optimal contact control with 30-day delay resulted in 720 deaths per 100,000 (at 25% isolation) and 510 deaths (at 50% isolation), losing to feedback control which held deaths to 620 and 250 respectively.
Endogenous and Exogenous Control of Virus-Host Dynamics · Georgia Tech
Considered and rejected
Considered and rejected: Rejected open-loop control using optimal steady-state bus voltages because it lacks feedback accuracy and cannot regulate against disturbances.
Networked DC microgrids control system for optimal power exchange with guaranteed stability · Imperial
Considered and rejected
Considered and rejected: Rejected linear quadratic regulator (LQR) open-loop control because it lacks input constraints and infinite-time convergence assumptions are unrealistic
Early Prediction of Adverse Clinical Events and Optimal Intervention in Intensive Care Units · JScholarship
Considered and rejected
Considered and rejected: Rejected open-loop constant molar flux control for active equilibration due to first-order drift in total internal pressure caused by environmental oxygen perturbations.
Developing in situ instrumentation to monitor anthropogenic change · Woods Hole
Considered and rejected
Considered and rejected: Open-loop tetanic activation control: rejected because sustaining open-loop tetanus caused rapid muscle deterioration and tissue damage, requiring closed-loop proxy evaluation instead.
Rapid wingbeats in insects through supra-resonant elasticity and actuation · Georgia Tech
Lost to a baseline
Controller 1 (feedforward shape propagation) was completely outperformed by feedback Controllers 2 and 3 under unknown obstacles and poor initial contact, achieving 0% success compared to 100% success for feedback controllers.
Snake and Snake Robot Locomotion in Complex, 3-D Terrain · JScholarship
Considered and rejected
Considered and rejected: Rejected parameterising predicted control inputs as open-loop perturbation sequences {c_0|k, ., c_{N-1|k}} with fixed observer and feedback gains because it provides poor performance and small feasible sets under packet loss.
Tried and failed
single-step open-loop dynamic optimization applied to nonlinear bioreactor trajectory optimization. Outcome: did not converge. Reason: numerical solver failed over full horizon, requiring receding horizon optimization instead
Tried and failed
offline open-loop optimal control without state feedback applied to satellite rendezvous docking under dynamic perturbations. Outcome: did not converge. Reason: lack of feedback prevented correction for dynamic perturbations during execution
Distributed and Time-Varying Optimization for Autonomy and Decision-Making · Georgia Tech
Tried and failed
iterative unrolled deep learning prediction without feedback applied to structural topology optimization. Outcome: worse than baseline. Reason: Uncoupled density sequence prediction failed to maintain volume fraction constraints compared to coupled physical feedback models.
Deep learning frameworks for structural topology optimization · Iowa State
Tried and failed
open-loop model-based predictive scheduling without local sensors applied to precision agricultural irrigation management. Outcome: worse than baseline. Reason: overestimated growth model coefficients and comparison against an experienced baseline operator
Considered and rejected
Considered and rejected: Rejected open-loop / batch experimental design when intermediate experimental outcomes can actively inform subsequent design decisions in non-parallel settings.
Accounting for Computational Expenditures in Bayesian Experimental Design · MIT
Closed-loop feedback is outperformed by simpler open-loop or feedforward strategies
Complex closed-loop feedback loops frequently introduce computational overhead, actuator saturation near zero crossings, or excessive resource consumption compared to simpler open-loop baselines. In several implementations, open-loop current control, correlation methods, or baseline schedules achieved lower tracking error, reduced costs, and avoided hardware complexity.
Lost to a baseline
Sequential Open-Loop Control (OLC) achieved higher steady-state tracking accuracy than MPC across several prediction horizons (Figure 5.4).
Guidance and Navigation Algorithms for Spacecraft Low-Thrust Proximity Operations: Formation Flight in Circular Relative Orbit · IRIS - POLITO - prod
Lost to a baseline
For process LDH 6 final state estimation, the open-loop reference model was more accurate than the direct soft-sensor and PF observer.
Novel soft-sensor applications and mechanistic models for biomanufacturing with Escherichia coli · DSpace-CRIS at TU Wien
Lost to a baseline
Open-loop system with rho_s = 0.5 achieved lower mean peak age (Delta = 4) than pipelined closed-loop feedback systems when feedback was heavily bottlenecked (mu_f < 0.6).
Age- and deviation-of-information in networked cyber-physical systems · Leibniz Universität Hannover Repository
Considered and rejected
Considered and rejected: Rejected open-loop/closed-loop compensation mechanisms requiring active control rod feedback in favor of open-loop moving-average detrended oscillations.
Analysis and Reproduction of Stainless-steel Integral Experiments Towards Nuclear Data Assimilation · EPFL
Considered and rejected
Considered and rejected: Rejected using closed-loop torque control with dedicated output torque sensing, choosing open-loop current control of quasi-direct drive actuators to reduce mass and mechanical complexity.
Enhancing Human Locomotion using a Robotic Knee Exoskeleton · Georgia Tech
Considered and rejected
Considered and rejected: Rejected closed-loop feedback control for breakthrough concentration generation due to complexity of required electronics and programming, choosing open-loop control with manual calibration instead
Characterisation of porous materials for the adsorption of volatile organic compounds · Imperial
Considered and rejected
Considered and rejected: Closed-loop gradient algorithm rejected because open-loop correlation requires fewer samples, half the multiplications per sampling interval, and provides better noise accuracy.
NON-ADAPTIVE ECHO CANCELLATION · HARVEST
Lost to a baseline
On BDD Day-Urban sequences, PID closed-loop sensor control achieved 0.75-0.76 recall vs 0.80 for open-loop baseline.
Closed Loop Perception for Resource Efficient Autonomous Systems · Georgia Tech
Lost to a baseline
Feedforward-enhanced feedback control had lower experimental performance (3.7% THD) compared to dynamic cancellation control (2.9% THD) due to duty-ratio saturation near zero crossings creating a flat dead-band
CONTROL OF HIGH-POWER-DENSITY LINE-INTERFACED POWER CONVERTERS · Cornell
Tried and failed
high-frequency feedback loops in the datapath applied to network congestion control. Outcome: worse than baseline. Reason: excessive CPU overhead from frequent feedback computation degraded throughput
Toward Predictable Networks · Cornell
Lost to a baseline
In the modular CZ framework under varying traffic rates, the feedback controller incurred higher total cost in CZ1 (36.96 vs 34.23) and fixed-speed flow control incurred higher overall total objective (76.77 vs 74.72) than the baseline without flow control.
Safety-critical optimal control in autonomous traffic systems · OpenBU
Lost to a baseline
In decreasing inertia perturbations, the closed-loop controller achieved a cost of 3.19, performing worse than the open-loop model baseline cost of 2.87.
Neuro-musculoskeletal Models: A Tool to Study the Contribution of Muscle Dynamics to Biological Motor Control · Publikationssystem UB Tuebingen
Excessive feedback gains and aggressive tuning induce instability and actuator saturation
High feedback gains, aggressive adaptation rates, and derivative terms frequently destabilize plants by triggering persistent oscillations or driving actuators into saturation. In nonlinear and time-varying applications, excessive feedback caused closed-loop divergence, limit cycles, or unintended systemic breakdowns.
Tried and failed
nested force feedback in position control loop applied to series elastic actuator position control. Outcome: worse than baseline. Reason: Inner force feedback degraded stability margins, forcing smaller position gains and worsening position tracking accuracy.
Towards versatile, high-performing and interactive humanoids · UT Austin
Tried and failed
fully automated closed-loop environmental control applied to building HVAC systems. Outcome: unstable. Reason: complex feedback dynamics produced unintended systemic breakdowns rather than stable autonomous regulation
Remote Control: Art, Technology, and the Politics of Distance (1966-1972) · MIT
Tried and failed
Aggressive convergence rate optimization in feedback control applied to nonlinear dynamical systems. Outcome: unstable. Reason: Excessively high feedback gains induced actuator saturation in nonlinear simulation despite theoretical linear certification
Safe online learning for nonlinear dynamical systems using control contraction metrics · Cranfield
Tried and failed
adaptive feedforward control with high learning rates applied to robot motion control with sensory feedback. Outcome: unstable. Reason: Excessively high adaptation rate in the feedforward loop caused dynamic control instability across trials.
Bio-inspired robotics: efference copies and adaptive feedforward control · Imperial
Tried and failed
derivative action in PID feedback control applied to piezoelectric actuator stabilization. Outcome: unstable. Reason: derivative gain consistently induced closed-loop oscillations regardless of parameter tuning
Strong Light — Matter Interactions in a Bow-Tie Cavity and Machine Learning-Enhanced Cooling · MIT
Tried and failed
open-loop to closed-loop current ramp-down switching applied to sensorless motor drive control. Outcome: unstable. Reason: Excessive ramp-down slope drops current prematurely before phase error converges, causing torque and speed oscillations.
Sensorless control of motoring/generating system for future more-electric turboprop aircraft · University of Nottingham Repository
Tried and failed
constant gain PID feedback control applied to time-varying acoustic emission regulation. Outcome: unstable. Reason: spatial density variations and temporal clearance kinetics caused control loop divergence
Microbubble Dynamics Monitoring and Control for Diagnosis and Treatment of Brain Cancer · Georgia Tech
Tried and failed
hardware-inferred memory delay line with feedback applied to real-time digital audio processing. Outcome: unstable. Reason: synthesis failed to infer block RAM and uncontrolled compounding feedback caused signal instability
Audio Processing on Field-Programmable Gate Arrays · Harvard
Considered and rejected
Considered and rejected: Rejected using unrealistically high control gains in simulation to bridge the stride length gap between open-loop and closed-loop because they fail on physical hardware.
TOWARD HIGH-PERFORMANCE SIMPLE MODELS OF LEGGED LOCOMOTION · Penn
Tried and failed
Concave nonlinear feedback control policy applied to Decentralized agent distribution control. Outcome: unstable. Reason: Low feedback exponents caused persistent switching oscillations and high steady-state error near the target distribution.
Effective task allocation frameworks for large-scale multiple agent systems. · Cranfield
Tried and failed
droop control with instantaneous vector discretization applied to grid-connected inverter power regulation. Outcome: unstable. Reason: positive feedback occurs when phase angle difference enters 90 to 270 degrees, causing active power divergence
Vector Based Control for Power Electronics Dominated AC Power Grid · Virginia Tech
Tried and failed
linearized state-feedback control applied to bistable nonlinear reaction-diffusion PDEs. Outcome: unstable. Reason: linearization failed to capture nonlinear dynamics, allowing trajectories to escape toward an alternate stable steady state
Control of agent-based models across scales · Imperial
Hardware latency and communication delays destabilize feedback loops
Time delays from sensor conditioning, communication buses, and digital processing introduce severe phase lag that triggers oscillations and limits closed-loop bandwidth. In multiple systems, feedback signals arriving with millisecond-scale latency degraded disturbance rejection or destabilized physical hardware.
Tried and failed
two-loop acceleration feedback autopilot applied to aerodynamically unstable vehicle control. Outcome: unstable. Reason: extreme sensitivity to actuator dynamics and lag causes closed-loop instability without rate-damping inner loops
Smart Projectile Performance Augmentation Using an Articulating Nose · Georgia Tech
Tried and failed
maximizing closed-loop control bandwidth applied to time-delayed opto-mechatronic feedback systems. Outcome: worse than baseline. Reason: Latency and phase lag caused severe noise amplification and degraded low-frequency disturbance rejection.
Opto-mechatronic disturbance compensation for free-space optical satellite communication · DSpace-CRIS at TU Wien
Considered and rejected
Considered and rejected: Feeding back measured ergometer position directly into real-time closed-loop simulation: rejected because 1-2 ms hardware latency destabilized the loop, requiring feedback of prescribed position instead.
Rapid wingbeats in insects through supra-resonant elasticity and actuation · Georgia Tech
Considered and rejected
Considered and rejected: Rejected continuous on-balance feedback measurements because finite lock-in response time and feedback loop lag caused excessive measurement slowdown and errors compared to off-balance acquisition.
The Electronic Compressibility of Rhombohedral Graphene Multilayers · MIT
Tried and failed
continuous sinusoidal feedback actuation applied to real-time aeroelastic vibration damping. Outcome: too slow. Reason: Update rate lag degraded phase synchronization accuracy and control response speed.
Considered and rejected
Considered and rejected: Rejected pure feedback control and teleoperation due to severe sensor/communication time delays and slow operational speed
Bio-Inspired Adaptive Control of Robotic Manipulators for Grasping in Orbital Space Missions · Carleton University Institutional Repository
Considered and rejected
Considered and rejected: Online maximum likelihood estimation (MLE) in real-time feedback loop rejected in favor of reactive extremum seeking control due to computational latency in the temporal control budget.
Tried and failed
coordinated feedback control with parameter tracking applied to distributed energy frequency regulation. Outcome: unstable. Reason: communication delays around 200 ms induce sustained oscillations through resonant mode amplification
Frequency Control of Virtual Power Plants · Research Repository UCD
Tried and failed
analog LVDT conditioning IC for position sensing applied to magnetic levitation closed-loop control. Outcome: unstable. Reason: signal processing IC introduced excessive phase lag around 100 Hz into the feedback loop
Bearingless Slice Motor and 6-DoF Position Sensor for Extracorporeal Pump · MIT
Lost to a baseline
Baseline strategy moving towards the intruder's azimuth angle achieved larger terminal separation ls than the optimal strategy in certain high-order dynamic trials due to cross-feedback computation delay.
Perception-Action Loop for Multi-Robot Systems with Deep Learning · Penn
Tried and failed
GPIB-controlled power supply for digital feedback loop applied to interferometer fringe locking. Outcome: unstable. Reason: Communication latency and quantized output steps caused discrete voltage jumps and signal disruptions.
Sensor noise and signal differentiation corrupt feedback and state estimation
Directly feeding back noisy measurements or numerically differentiating latency-affected signals introduces high-frequency actuator chatter and estimator degradation. Closed-loop parameter identification also becomes biased and inconsistent when measurement noise correlates with control inputs under misspecified noise models.
Lost to a baseline
In the introductory benchmark with measurement noise disturbance n(t) at t = 7 s, H-infinity reached a peak amplitude of 1.341 and recovered in 98.86 ms, whereas Polynomial control took 996.5 ms (reaching amplitude 1.338) and State Feedback took 999 ms (reaching amplitude 1.802).
Analisi e paragone di diversi concetti di controllo per un pendolo inverso con un modo di risonanza · SUPSI - ARIS
Tried and failed
direct force sensor feedback control applied to haptic robot manipulator interaction. Outcome: unstable. Reason: sensor noise in the closed-loop force feedback severely degraded control stability
Virtual mechanism with feedback force · Iowa State
Considered and rejected
Considered and rejected: Rejected numerical differentiation of slow-sampled, latency-affected exteroceptive pose signals for floating-base velocity feedback due to high-frequency actuator chatter and instability.
Employing symmetry in dynamics and motion control of robotic mechanisms · DSpace-CRIS at TU Wien
Considered and rejected
Considered and rejected: Rejected full-state feedback control for physical implementations due to degradation/instability from measurement noise, delay, derivative noise, and signal saturation.
Design Methodologies for Simple Adaptive Controllers with Applications to Spacecraft Proximity Operations · Carleton University Institutional Repository
Tried and failed
Direct identification under feedback applied to closed-loop dynamic systems. Reason: Noise-input correlation under misspecified noise models produces biased, inconsistent parameter estimates
Learning and Control with Generalization and Stability Certificates · EPFL
Considered and rejected
Considered and rejected: Rejected direct numerical differentiation of the cantilever deflection signal for ORT velocity feedback due to amplified noise and estimator computational latency, choosing a fixed time-delay instead.
Data-Driven Methods for Controller Design in Atomic Force Microscopy · EPFL
Open-loop parameter tuning and training fail during closed-loop execution
Optimizing models or control trajectories solely on open-loop data fails to account for downstream plant dynamics and recurrent feedback interactions. When deployed in closed-loop settings, these open-loop calibrated controllers accumulate integration errors and induce severe oscillatory behavior.
Tried and failed
piecewise-affine autoregressive modeling in predictive control applied to building HVAC frequency response control. Outcome: did not generalise. Reason: superior open-loop prediction accuracy did not translate to improved closed-loop tracking performance
Applications of Data-driven Predictive Control to Building Energy Systems · EPFL
Tried and failed
tuning MPC using open-loop trajectories applied to legged robot closed-loop control. Outcome: did not generalise. Reason: approximate discrete-time dynamics compounded integration errors like foot penetration and slip during closed-loop execution
Tailoring Complexity of Model-Based Controllers for Legged Robots · MIT
Tried and failed
open-loop parameter calibration excluding downstream plant dynamics applied to actuator control driven by biosignals. Outcome: unstable. Reason: omitting plant dynamics and feedback controller during optimization caused severe oscillatory behavior during closed-loop execution
Toward impedance control in human-machine interfaces for upper-limb prostheses · Imperial
Considered and rejected
Considered and rejected: Rejected pure open-loop training dataset evaluation for LSTM building dynamics in favor of closed-loop multi-step autoregressive rollout evaluation to test error accumulation and stability
Deep Reinforcement Learning-based Control Strategies for Enhancing Energy Management in HVAC Systems · IRIS - POLITO - prod
Tried and failed
static input balance tuned in isolated units applied to recurrently connected network simulations. Outcome: unstable. Reason: isolated tuning failed to account for recurrent network feedback dynamics
Pure feedback control fails without feedforward or predictive augmentation
Relying strictly on reactive feedback without feedforward terms or integral action leaves persistent steady-state tracking errors and poor dynamic compensation. Non-predictive feedback controllers exhibited inferior reference tracking compared to two-degree-of-freedom and predictive architectures.
Lost to a baseline
Omitting linear feedback gain adaptation (ˆkx adaptation turned off, γx = 0) resulted in increased oscillations and worse tracking error performance compared to adapting linear feedback gains.
Robust and Model Reference Adaptive Control for Nonlinear Uncertainty in Native Space · Virginia Tech
Considered and rejected
Considered and rejected: Rejected pure feedback PID low-level SEA control because human tissue damping and dynamic back-driving degraded tracking without feedforward compensation.
Adaptive User State Estimation for Assisting Human Locomotion Using Powered Hip Exoskeletons · Georgia Tech
Lost to a baseline
PD trajectory tracking under feedback-only control exhibited an average steady-state tracking error of e_avg = 0.0193 m due to lack of feed-forward or integral terms.
Multi-Objective Control for Physical and Cognitive Human-Exoskeleton Interaction · Virginia Tech
Lost to a baseline
Non-predictive controllers (PI and state feedback) achieved lower determination coefficients (R2 = 0.613 and R2 = 0.857) than the proposed 2DoF controller (R2 = 0.947) and MPC (R2 = 0.987) for NOx output tracking without input saturation.
A multi-level cooperative strategy for reducing fuel consumption and transient pollutant emissions of hybrid electric vehicles · DSpace-CRIS at TU Wien
Left open by the authors
Problems the authors named and did not get to.
Left open
Integrate online analytical sensors with continuous-flow reactors to enable real-time closed-loop feedback control of nanomaterial synthesis parameters. Blocker: Requires a physical continuous-flow reactor setup and integrated online analytical hardware.
Scalable Synthesis of Nanomaterials in Continuous-Flow Reactors · Georgia Tech
Left open
Develop a closed-loop feedback system controlling heating and cooling rates or thermal training for repeatable SMA actuation strain in knitted textiles. Blocker: Requires the physical knitted textile apparatus, Nitinol SMA actuators, sensors, and thermal control hardware
Left open
Implement closed-loop feedback or robust model predictive control under uncertainty for the simulated cellulose oxidation reactor. Blocker: None
Set-Based Fault Diagnosis and Dynamic Optimization of a Cellulose Oxidation Reactor · Georgia Tech
Left open
Perform closed-loop conditional auditory feedback with a 10 ms window across multiple discrete feedback delays to validate the synaptic eligibility trace shape. Blocker: Requires live zebra finches, chronic Neuropixels 2.0 neural implants, and specialized closed-loop electrophysiology wet-lab hardware.
Driving Temporally Precise Learning in Individual Premotor Neurons using Closed-Loop Neurofeedback · MIT
Left open
Apply the Limited Individual Lifetime Differential Evolution algorithm to optimize experimental Bose-Einstein condensate cycle parameters and cycle time. Blocker: Requires direct feedback and closed-loop control on the physical cold-atom experimental apparatus.
Experimental setup for fast BEC generation and number-stabilized atomic ensembles · Leibniz Universität Hannover Repository
Left open
Integrate high-level planning with a reusable atomic skill library to handle noisy execution feedback and compounding real-world errors in robotic manipulation. Blocker: Requires a physical robot manipulation setup and environment to evaluate closed-loop real-world error recovery.
Leveraging Pretrained Vision and Language Models for Robotic Manipulation · Georgia Tech
Left open
Test timing-specified robotic handover controllers with corrective feedback mechanisms in real-world industrial environments with human workers. Blocker: Requires physical industrial robotic manipulators and access to real industrial operating environments.
Robot Controllers, Gaze Behaviors and Human Motion Datasets for Object Handovers · Cornell
Left open
Simulate bimanual haptic feedback within motor control models to optimize feedback gains and modalities. Blocker: Specific motor control model formulations and objective functions for optimization are not detailed
Bimanual coordination classification, influence, and haptic augmentation for human-operated robotic systems · UT Austin
Left open
Develop and implement adaptive Smith-predictor-based feedback controllers for improved setpoint regulation in rolling-diaphragm hydrostatic transmission actuators. Blocker: Requires the physical hydrostatic transmission robotic hardware or proprietary experimental setup for validation
Rolling-diaphragm hydrostatic transmission for the remote actuation of high-performance robots · IRIS - UNITN - prod
Left open
Extend the causal discovery architecture to model cyclic feedback relationships and time-resolved dynamic systems via steady-states or stochastic differential equations. Blocker: None
Practical Algorithms for Modeling Causality to Accelerate Scientific Discovery · MIT
Checking a claim in this area?
We can run the same search on any method or claim. If nothing turns up, we will say so, and that proves nothing on its own.