RGRavi Gondhalekar

Ravi Gondhalekar

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Ravi Gondhalekar has a total of 102 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Tube-Based Model Predictive Control with Uncertainty Identification for Autonomous Spacecraft Maneuvers

    by Charles E. Oestreich, Richard Linares, Ravi Gondhalekar

    Autonomous spacecraft must be robust toward uncertainty and disturbances in the system while achieving required levels of performance. As an example, robotic servicing spacecraft must intercept target objects with potentially unknown dynamic properties during active debris removal efforts. This work presents tube-based model predictive control (MPC) with uncertainty identification as a strategy to enhance performance while maintaining robustness in autonomous maneuvers. The proposed algorithm, which is an extension of the standard tube-based MPC framework, measures and predicts the exogenous input to the system (i.e., the uncertainty) online. This in turn enables the robust tube to be shrunk and grown appropriately as the trajectory progresses in both predicted time and actual time. The algorithm is demonstrated in a simulated intercept maneuver with a tumbling target whose inertia tensor is uncertain. Results indicate two improvements over the standard tube-based algorithm: first, better performance is obtained when the initial exogenous input bounds are overconservative, and, second, there is greater flexibility in encouraging robustness when the exogenous input bounds are overly optimistic since the robust tube is updated online. As such, tube-based MPC with uncertainty identification represents an incremental step in enhancing the flexibility of autonomous spacecraft in addressing uncertain scenarios and environments.


  • Autonomous Six-Degree-of-Freedom Spacecraft Docking with Rotating Targets via Reinforcement Learning

    by Charles E. Oestreich, Richard Linares, Ravi Gondhalekar

    A policy for six-degree-of-freedom docking maneuvers with rotating targets is developed through reinforcement learning and implemented as a feedback control law. Potential clients for satellite servicing and orbital debris objects are often rotating around a constant axis within their respective Earth orbits. In the context of such missions, reinforcement learning provides an appealing framework for robust, autonomous maneuvers in uncertain environments with low on-board computational cost. This work uses proximal policy optimization to produce a docking policy for rotating or nonrotating targets that is valid over a portion of the six-degree-of-freedom state space while striving to minimize performance and control costs. Experiments using the simulated Apollo transposition and docking maneuver with an induced spin in the lunar module exhibit the policy’s capabilities and provide a comparison with standard optimal control techniques. Furthermore, specific challenges and workarounds, as well as a discussion on the benefits and disadvantages of reinforcement learning for docking policies, are discussed to facilitate future research. As such, this work will serve as a foundation for further investigation of learning-based control laws for spacecraft proximity operations in uncertain environments.


  • Real-Time Detection of Infusion Site Failures in a Closed-Loop Artificial Pancreas

    by Daniel P. Howsmon, Nihat Baysal, Bruce A. Buckingham, Gregory P. Forlenza, Trang T. Ly, David M. Maahs, Tatiana Marcal, Lindsey Towers, Eric Mauritzen, Sunil Deshpande, et al.

    As evidence emerges that artificial pancreas systems improve clinical outcomes for patients with type 1 diabetes, the burden of this disease will hopefully begin to be alleviated for many patients and caregivers. However, reliance on automated insulin delivery potentially means patients will be slower to act when devices stop functioning appropriately. One such scenario involves an insulin infusion site failure, where the insulin that is recorded as delivered fails to affect the patient’s glucose as expected. Alerting patients to these events in real time would potentially reduce hyperglycemia and ketosis associated with infusion site failures. An infusion site failure detection algorithm was deployed in a randomized crossover study with artificial pancreas and sensor-augmented pump arms in an outpatient setting. Each arm lasted two weeks. Nineteen participants wore infusion sets for up to 7 days. Clinicians contacted patients to confirm infusion site failures detected by the algorithm and instructed on set replacement if failure was confirmed. In real time and under zone model predictive control, the infusion site failure detection algorithm achieved a sensitivity of 88.0% (n = 25) while issuing only 0.22 false positives per day, compared with a sensitivity of 73.3% (n = 15) and 0.27 false positives per day in the SAP arm (as indicated by retrospective analysis). No association between intervention strategy and duration of infusion sets was observed (P = .58). As patient burden is reduced by each generation of advanced diabetes technology, fault detection algorithms will help ensure that patients are alerted when they need to manually intervene. www.clinicaltrials.gov,NCT02773875


  • Simultaneous model predictive control and moving horizon estimation for blood glucose regulation in Type 1 diabetes

    by David A. Copp, Ravi Gondhalekar, João P. Hespanha

    Summary: A new estimation and control approach for the feedback control of an artificial pancreas to treat Type 1 diabetes mellitus is proposed. In particular, we present a new output‐feedback predictive control approach that simultaneously solves the state estimation and control objectives by means of a single min‐max optimization problem. This involves optimizing a cost function with both finite forward and backward horizons with respect to the unknown initial state, unmeasured disturbances and noise, and future control inputs and is similar to simultaneously solving a model predictive control (MPC) problem and a moving horizon estimation (MHE) problem. We incorporate a novel asymmetric output cost to penalize dangerous low blood glucose values more severely than less harmful high blood glucose values. We compare this combined MPC/MHE approach to a control strategy that uses state‐feedback MPC preceded by a Luenberger observer for state estimation. In‐silico results showcase several advantages of this new simultaneous MPC/MHE approach, including fewer hypoglycemic events without increasing the number of hyperglycemic events, faster insulin delivery in response to a meal consumption, and shorter insulin pump suspensions, resulting in smoother blood glucose trajectories.


  • Extremum Seeking Control for Personalized Zone Adaptation in Model Predictive Control for Type 1 Diabetes

    by Zhixing Cao, Ravi Gondhalekar, Eyal Dassau, Francis J. Doyle

    Zone model predictive control has proven to be an effective closed-loop method to regulate blood glucose for people with type 1 diabetes (T1D). In this paper, we present a universal model-free optimization scheme for adapting the zone for T1D patients individually. The adaptation is based on a clinical glycemic risk index named relative regularized glycemic penalty index (rrGPI), which is calculated from glucose measurements by a continuous glucose monitor. The scheme's objective is to minimize rrGPI by simultaneously modulating a controller's blood glucose target zone's upper bound and lower bound. The adaptation mechanism is based on extremum seeking control, in which the zone boundaries are driven by gradient estimation obtained by continuously sinusoidally modulating and demodulating the rrGPI readings. To improve the adaptation method's robustness against uncertainties, a decaying feedback gain and a vanishing dither signal are employed. in-silico trials suggested that the personalized optimized zone can be reached within a week of adaptation. Both for announced and unannounced meals, the proposed method outperforms the fixed zone [80, 140] mg/dL, which has been employed in the authors' clinical trials. It is also shown that the developed method has strong robustness against real-life uncertainties.


  • Twelve-Week 24/7 Ambulatory Artificial Pancreas With Weekly Adaptation of Insulin Delivery Settings: Effect on Hemoglobin A1c and Hypoglycemia

    by Eyal Dassau, Jordan E. Pinsker, Yogish C. Kudva, Sue A. Brown, Ravi Gondhalekar, Chiara Dalla Man, Steve Patek, Michele Schiavon, Vikash Dadlani, Isuru Dasanayake, et al.

    OBJECTIVE Artificial pancreas (AP) systems are best positioned for optimal treatment of type 1 diabetes (T1D) and are currently being tested in outpatient clinical trials. Our consortium developed and tested a novel adaptive AP in an outpatient, single-arm, uncontrolled multicenter clinical trial lasting 12 weeks. RESEARCH DESIGN AND METHODS Thirty adults with T1D completed a continuous glucose monitor (CGM)-augmented 1-week sensor-augmented pump (SAP) period. After the AP was started, basal insulin delivery settings used by the AP for initialization were adapted weekly, and carbohydrate ratios were adapted every 4 weeks by an algorithm running on a cloud-based server, with automatic data upload from devices. Adaptations were reviewed by expert study clinicians and patients. The primary end point was change in hemoglobin A1c (HbA1c). Outcomes are reported adhering to consensus recommendations on reporting of AP trials. RESULTS Twenty-nine patients completed the trial. HbA1c, 7.0 ± 0.8% at the start of AP use, improved to 6.7 ± 0.6% after 12 weeks (−0.3, 95% CI −0.5 to −0.2, P < 0.001). Compared with the SAP run-in, CGM time spent in the hypoglycemic range improved during the day from 5.0 to 1.9% (−3.1, 95% CI −4.1 to −2.1, P < 0.001) and overnight from 4.1 to 1.1% (−3.1, 95% CI −4.2 to −1.9, P < 0.001). Whereas carbohydrate ratios were adapted to a larger extent initially with minimal changes thereafter, basal insulin was adapted throughout. Approximately 10% of adaptation recommendations were manually overridden. There were no protocol-related serious adverse events. CONCLUSIONS Use of our novel adaptive AP yielded significant reductions in HbA1c and hypoglycemia.


  • Extremum Seeking Control Based Zone Adaptation for Zone Model Predictive Control in Type 1 Diabetes

    by Zhixing Cao, Eyal Dassau, Ravi Gondhalekar, Francis J. Doyle Iii