profile photo

Devansh Agrawal

I am a final year PhD student in Aerospace Engineering and Robotics at the University of Michigan, advised by Professor Dimitra Panagou. My research interests include the synthesis and analysis of safety critical control and perception systems for nonlinear robotic systems. My research lies at the intersection of theory and practice, where I develop novel theoretical results to characterize the performance and behavior of control algorithms, and then demonstrate and validate them experimentally using quadrotors and ground rovers. Most of my code is open-sourced.

I did my undergrad from Imperial College London, where I earned my BEng and MEng in Aeronautical Engineering. I was awared the Royal Aeronautical Society Prize, and was on the Dean's List all four years. My masters thesis (pdf) was on the control of extremely flexible structures. During my undergrad I co-founded the Imperial Drone Society, and the Rocketry Team.

I also spent a year at Massachussets Institute of Technology, as an exchange student during my third year. During this time I worked on my projects, including the design of a super-short takeoff and landing vehicle, the optimization of silent-properless aircrafts, and was actively involved in the MIT Rocket Team.

Visit my tech-notes for some technical notes that I have written up. If you find them useful (or find mistakes), please let me know!

I am always interested in technically challenging projects, and love to discuss ideas with people. Feel free to reach out if you would like to chat.

Email  /  GitHub  /  Google Scholar  /  LinkedIn

Publications

1. Multi-Agent Clarity-Aware Dynamic Coverage with Gaussian Processes

Devansh Agrawal and Dimitra Panagou
IEEE CDC 2024
2024
arxiv / pdf / code / /

This paper presents two algorithms for multi-agent dynamic coverage in spatiotemporal environments, where the coverage algorithms are informed by the method of data assimilation. In particular, we show that by explicitly modeling the environment using a Gaussian Process (GP) model, and considering the sensing capabilities and the dynamics of a team of robots, we can design an estimation algorithm and multi-agent coverage controller that explores and estimates the state of the spatiotemporal environment. The uncertainty of the estimate is quantified using clarity, an information-theoretic metric, where higher clarity corresponds to lower uncertainty. By exploiting the relationship between GPs and Stochastic Differential Equations (SDEs) we quantify the increase in clarity of the estimated state at any position due to a measurement taken from any other position. We use this relationship to design two new coverage controllers, both of which scale well with the number of agents exploring the domain, assuming the robots can share the map of the clarity over the spatial domain via communication. We demonstrate the algorithms through a realistic simulation of a team of robots collecting wind data over a region in Austria.

2. gatekeeper: Online Safety Verification and Control for Nonlinear Systems in Dynamic Environments

Devansh Agrawal, Ruichang Chen, and Dimitra Panagou
IEEE T-RO 2024
2024
arxiv / pdf / video / code / link / / /

@ARTICLE{10665919,
  author={Agrawal, Devansh Ramgopal and Chen, Ruichang and Panagou, Dimitra},
  journal={IEEE Transactions on Robotics},
  title={gatekeeper: Online Safety Verification and Control for Nonlinear Systems in Dynamic Environments},
  year={2024},
  volume={},
  number={},
  pages={1-17},
  keywords={Trajectory;Safety;Robots;Logic gates;Robot sensing systems;Nonlinear dynamical systems;Quadrotors;Collision Avoidance;Motion and Path Planning;Aerial Systems: Applications;Safety-Critical Control},
  doi={10.1109/TRO.2024.3454415}
}

This paper presents the gatekeeper algorithm, a real-time and computationally-lightweight method that ensures that trajectories of a nonlinear system satisfy safety constraints despite sensing limitations. gatekeeper integrates with existing path planners and feedback controllers by introducing an additional verification step to ensure that proposed trajectories can be executed safely, despite nonlinear dynamics subject to bounded disturbances, input constraints and partial knowledge of the environment. Our key contribution is that (A) we propose an algorithm to recursively construct safe trajectories by numerically forward propagating the system over a (short) finite horizon, and (B) we prove that tracking such a trajectory ensures the system remains safe for all future time, i.e., beyond the finite horizon. We demonstrate the method in a simulation of a dynamic firefighting mission, and in physical experiments of a quadrotor navigating in an obstacle environment that is sensed online. We also provide comparisons against the state-of-the-art techniques for similar problems.

3. Eclares: Energy-Aware Clarity-Driven Ergodic Search

Kaleb Ben Naveed, Devansh Agrawal, Christopher Vermillion, Dimitra Panagou
IEEE ICRA 2024
2024
arxiv / pdf / video / code / link / / /

@inproceedings{naveed2024eclares,
  title={Eclares: Energy-aware clarity-driven ergodic search},
  author={Naveed, Kaleb Ben and Agrawal, Devansh and Vermillion, Christopher and Panagou, Dimitra},
  booktitle={2024 IEEE International Conference on Robotics and Automation (ICRA)},
  pages={14326--14332},
  year={2024},
  organization={IEEE}
}

Planning informative trajectories while considering the spatial distribution of the information over the environment, as well as constraints such as the robot’s limited battery capacity, makes the long-time horizon persistent coverage problem complex. Ergodic search methods consider the spatial distribution of environmental information while optimizing robot trajectories; however, current methods lack the ability to construct the target information spatial distribution for environments that vary stochastically across space and time. Moreover, current coverage methods dealing with battery capacity constraints either assume simple robot and battery models or are computationally expensive. To address these problems, we propose a framework called Eclares, in which our contribution is two-fold. 1) First, we propose a method to construct the target information spatial distribution for ergodic trajectory optimization using clarity, an information measure bounded between [0, 1]. The clarity dynamics allow us to capture information decay due to a lack of measurements and to quantify the maximum attainable information in stochastic spatiotemporal environments. 2) Second, instead of directly tracking the ergodic trajectory, we introduce the energy-aware (eware) filter, which iteratively validates the ergodic trajectory to ensure that the robot has enough energy to return to the charging station when needed. The proposed eware filter is applicable to nonlinear robot models and is computationally lightweight. We demonstrate the working of the framework through a simulation case study.

4. Advances in the Theory of Control Barrier Functions: Addressing practical challenges in safe control synthesis for autonomous and robotic systems

Kunal Garg, James Usevitch, Joseph Breeden, Mitchell Black, Devansh Agrawal, Hardik Parwana, Dimitra Panagou
Annual Reviews in Control
2024
pdf / link / / /

@article{garg2024advances,
  title={Advances in the Theory of Control Barrier Functions: Addressing practical challenges in safe control synthesis for autonomous and robotic systems},
  author={Garg, Kunal and Usevitch, James and Breeden, Joseph and Black, Mitchell and Agrawal, Devansh and Parwana, Hardik and Panagou, Dimitra},
  journal={Annual Reviews in Control},
  volume={57},
  pages={100945},
  year={2024},
  publisher={Elsevier}
}

This tutorial paper presents recent work of the authors that extends the theory of Control Barrier Functions (CBFs) to address practical challenges in the synthesis of safe controllers for autonomous systems and robots. We present novel CBFs and methods that handle safety constraints (i) with time and input constraints under disturbances, (ii) with high-relative degree under disturbances and input constraints, and (iii) that are affected by adversarial inputs and sampled-data effects. We then present novel CBFs and adaptation methods that prevent loss of validity of the CBF, as well as methods to tune the parameters of the CBF online to reduce conservatism in the system response. We also address the pointwise-only optimal character of CBF-induced control inputs by introducing a CBF formulation that accounts for future trajectories, as well as implementation challenges such as how to preserve safety when using output feedback control and zero-order-hold control. Finally we consider how to synthesize non-smooth CBFs when discontinuous inputs and multiple constraints are present.

5. gatekeeper: Online safety verification and control for nonlinear systems in dynamic environments

Devansh Agrawal, Ruichang Chen, Dimitra Panagou
IEEE IROS 2023
2023
arxiv / pdf / video / code / link / / /

@inproceedings{agrawal2023gatekeeper,
  title={gatekeeper: Online safety verification and control for nonlinear systems in dynamic environments},
  author={Agrawal, Devansh and Chen, Ruichang and Panagou, Dimitra},
  booktitle={2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)},
  pages={259--266},
  year={2023},
  organization={IEEE}
}

This paper presents the gatekeeper algorithm, a real-time and computationally-lightweight method to ensure that nonlinear systems can operate safely in dynamic environments despite limited perception. gatekeeper integrates with existing path planners and feedback controllers by introducing an additional verification step that ensures that proposed trajectories can be executed safely, despite nonlinear dynamics subject to bounded disturbances, input constraints and partial knowledge of the environment. Our key contribution is that (A) we propose an algorithm to recursively construct committed trajectories, and (B) we prove that tracking the committed trajectory ensures the system is safe for all time into the future. The method is demonstrated on a complicated firefighting mission in a dynamic environment, and compares against the state-of-the-art techniques for similar problems.

6. Sensor-based Planning and Control for Robotic Systems: Introducing Clarity and Perceivability

Devansh Agrawal and Dimitra Panagou
IEEE L-CSS and CDC 2023
2023
pdf / link / / /

@article{agrawal2023sensor,
  title={Sensor-based planning and control for robotic systems: Introducing clarity and perceivability},
  author={Agrawal, Devansh R and Panagou, Dimitra},
  journal={IEEE Control Systems Letters},
  year={2023},
  publisher={IEEE}
}

In this letter, we first introduce an information measure, termed clarity , motivated by information entropy, and show that it has intuitive properties relevant to dynamic coverage control and informative path planning. Clarity defines on a scale of [0,1] the quality of the information that we have about a variable of interest in an environment. Clarity lower bounds the expected estimation error of any estimator, and is used as the information metric in the notion of perceivability , which is defined later on and is the primary contribution of this letter. Perceivability captures whether a given robotic (or more generally, sensing and control) system has sufficient sensing and actuation capabilities to gather desired information about an environment. We show that perceivability relates to the reachability of an augmented system, which encompasses the robot dynamics and the clarity about the environment, and we derive the corresponding Hamilton-Jacobi-Bellman equations. Thus, we provide an algorithm to measure an environment’s perceivability, and obtain optimal controllers that maximize information gain. In simulations, we demonstrate how clarity is a useful concept for planning trajectories, how perceivability can be determined using reachability analysis, and how a Control Barrier Function controller can be used to design controllers to maintain a desired level of information.

7. Safe and robust observer-controller synthesis using control barrier functions

Devansh Agrawal, Dimitra Panagou
IEEE L-CSS and CDC 2022
2022
pdf / video / video 2 / link / / /

@article{agrawal2022safe,
  title={Safe and robust observer-controller synthesis using control barrier functions},
  author={Agrawal, Devansh R and Panagou, Dimitra},
  journal={IEEE Control Systems Letters},
  volume={7},
  pages={127--132},
  year={2022},
  publisher={IEEE}
}

This letter addresses the synthesis of safety-critical controllers using estimate feedback. We propose an observer-controller interconnection to ensure that the nonlinear system remains safe despite bounded disturbances on the system dynamics and measurements that correspond to partial state information. The co-design of observers and controllers is critical, since even in undisturbed cases, observers and controllers designed independently may not render the system safe. We propose two approaches to synthesize observer-controller interconnections. The first approach utilizes Input-to-State Stable observers, and the second uses Bounded Error observers. Using these stability and boundedness properties of the observation error, we construct novel Control Barrier Functions that impose inequality constraints on the control inputs which, when satisfied, certifies safety. We propose quadratic program-based controllers to satisfy these constraints, and prove Lipschitz continuity of the derived controllers. Simulations and experiments on a quadrotor demonstrate the efficacy of the proposed methods.

8. A Constructive Method for Designing Safe Multirate Controllers for Differentially-Flat Systems

Devansh Agrawal, Hardik Parwana, Ryan K Cosner, Ugo Rosolia, Aaron D Ames, Dimitra Panagou
IEEE L-CSS and ACC 2022
2021
pdf / video / link / / /

@ARTICLE{9655322,
  author={Agrawal, Devansh R. and Parwana, Hardik and Cosner, Ryan K. and Rosolia, Ugo and Ames, Aaron D. and Panagou, Dimitra},
  journal={IEEE Control Systems Letters},
  title={A Constructive Method for Designing Safe Multirate Controllers for Differentially-Flat Systems},
  year={2022},
  volume={6},
  number={},
  pages={2138-2143},
  doi={10.1109/LCSYS.2021.3136465}
}

This paper introduces the notion of an Input Constrained Control Barrier Function (ICCBF), as a method to synthesize safety-critical controllers for nonlinear control-affine systems with input constraints. The method identifies a subset of the safe set of states, and constructs a controller to render the subset forward invariant. The feedback controller is represented as the solution to a quadratic program, which can be solved efficiently for real-time implementation. Furthermore, we show that ICCBFs are a generalization of Higher Order Control Barrier Functions, and thus are applicable to systems of non- uniform relative degree. Simulation results are presented for the adaptive cruise control problem, and a spacecraft rendezvous problem.

9. Safe Control Synthesis via Input Constrained Control Barrier Functions

Devansh Agrawal and Dimitra Panagou
IEEE CDC
2021
pdf / video / link / / /

@INPROCEEDINGS{9682938,
  author={Agrawal, Devansh R. and Panagou, Dimitra},
  booktitle={2021 60th IEEE Conference on Decision and Control (CDC)},
  title={Safe Control Synthesis via Input Constrained Control Barrier Functions},
  year={2021},
  volume={},
  number={},
  pages={6113-6118},
  doi={10.1109/CDC45484.2021.9682938}
}

This paper introduces the notion of an Input Constrained Control Barrier Function (ICCBF), as a method to synthesize safety-critical controllers for nonlinear control-affine systems with input constraints. The method identifies a subset of the safe set of states, and constructs a controller to render the subset forward invariant. The feedback controller is represented as the solution to a quadratic program, which can be solved efficiently for real-time implementation. Furthermore, we show that ICCBFs are a generalization of Higher Order Control Barrier Functions, and thus are applicable to systems of non- uniform relative degree. Simulation results are presented for the adaptive cruise control problem, and a spacecraft rendezvous problem.

10. Mind The Gap: Real-time Decentralized Distance Estimation using Ultrasound and Bluetooth across Multiple Smartphones

Devansh Agrawal, Peter Lyon, Martin Frobisher, Andy Doherty, Ben Allen, Freddie Rawlins
ArXiv
2020
pdf / link / / /

@article{agrawal2020mind,
  title={Mind The Gap: Real-time Decentralized Distance Estimation using Ultrasound and Bluetooth across Multiple Smartphones},
  author={Agrawal, Devansh R and Lyon, Peter and Frobisher, Martin and Doherty, Andy and Allen, Ben and Rawlins, Freddie},
  journal={arXiv preprint arXiv:2008.13564},
  year={2020}
}

Robust, low-cost solutions are needed to maintain social distancing guidelines during the COVID-19 pandemic. We establish a method to measure the distance between multiple phones across a large number of closely spaced smartphones with a median absolute error of 8.5 cm. The application works in real-time, using Time of Flight of near-ultrasound signals, providing alerts with sufficient responsiveness to be useful for distancing while devices are in users pockets and they are moving at walking speed. The approach is decentralized, requires no additional hardware, and can operate in the background without an internet connection. We have no device specific requirements nor need any manual calibration or device synchronization. It has been tested with over 20 different phones models, from both the Android and iOS systems in the past 5 years. To the best of our knowledge, this is the first successful such implementation, and has 25000 users at time of publishing.

11. Higher thrust-to-power with large electrode gap spacing electroaerodynamic devices for aircraft propulsion

Haofeng Xu, Nicolas Gomez-Vega, Devansh Agrawal and Steven R H Barrett
Journal of Physics D: Applied Physics
2019
pdf / link / / /

@article{Xu_2019,
	doi = {10.1088/1361-6463/ab4a4c},
	url = {https://doi.org/10.1088/1361-6463/ab4a4c},
	year = 2019,
	month = {oct},
	publisher = {{IOP} Publishing},
	volume = {53},
	number = {2},
	pages = {025202},
	author = {Haofeng Xu and Nicolas Gomez-Vega and Devansh R Agrawal and Steven R H Barrett},
	title = {Higher thrust-to-power with large electrode gap spacing electroaerodynamic devices for aircraft propulsion},
	journal = {Journal of Physics D: Applied Physics},
}

Electroaerodynamic (EAD) devices, which produce a propulsive force in air by electrostatic acceleration, have been demonstrated as a method of propulsion for airplanes. However, achieving sufficient thrust-to-power is a significant challenge in developing EAD aircraft which are practical. Theory predicts that devices with larger inter-electrode gap spacing will enable higher thrust-to-power, but most experimental work has been limited to gap spacings of less than 80 mm. Those studies which have investigated spacings of greater than 100 mm have found results deviating from theory, with lower thrust-to-power than predicted. We performed experiments between 50 and 300 mm gap spacing and conclude that three effects explain the discrepancy: ‘leakage current’ from the electrodes to the surroundings, which does not produce thrust but increases measured electrical power; reverse corona emission from the collecting electrode, which reduces thrust and increases power; and the electric potential of the thruster relative to its surroundings, which affects both leakage current and reverse corona emission. Our results show that if these effects are accounted for, the existing EAD theory is correct without modification beyond its previous range of validity and is applicable to wire-to-cylinder EAD devices up to at least 300 mm gap spacing. We support our experimental results with two-dimensional numerical simulations, which show that the experimental current and thrust, including effects of leakage current, can be reproduced by computation with 12% error—an important step towards numerical design and optimization. By experimentally replicating equilibrium in-flight conditions, we measure thrust-to-power in the laboratory of up to 15 N kW−1 for large gap spacing thrusters at practically useful thrust levels. This is two to three times higher than current implementations with smaller gap spacings, suggesting that large gap spacing thrusters will be suitable for future EAD-propelled flight applications at thrust-to-power competitive with or exceeding conventional propulsion.

12. Wind Tunnel Testing of a Blown Flap Wing

Devansh Agrawal, Faisal Asad, Blake M. Berk, Trevor Long, Jackson Lubin, Christopher Courtin, Mark Drela, R John Hansman and Jacqueline L. Thomas
AIAA Aviation Forum 2019
2019
pdf / link / / /

@inbook{doi:10.2514/6.2019-3170,
  author = {Devansh Agrawal and Faisal Asad and Blake M. Berk and Trevor Long and Jackson Lubin and Christopher Courtin and Mark Drela and R John Hansman and Jacqueline L. Thomas},
  title = {Wind Tunnel Testing of a Blown Flap Wing},
  booktitle = {AIAA Aviation 2019 Forum},
  chapter = {},
  pages = {},
  doi = {10.2514/6.2019-3170},
  URL = {https://arc.aiaa.org/doi/abs/10.2514/6.2019-3170},
  eprint = {https://arc.aiaa.org/doi/pdf/10.2514/6.2019-3170}
}

This paper presents wind tunnel measurements of blown flapped airfoil performance forapplication to distributed electric propulsion STOL aircraft. The 2D airfoil wind tunnelmodel features a simple slotted flap, and closely-spaced spanwise-distributed propellers drivenby electric motors. Measurements of lift, pitching moment and net streamwise force (dragminus thrust) were made over a range of propeller RPM, angle of attack and flap angle. Liftcoefficients up to 9 were measured for practical blowing levels. High lift was also measuredwith net streamwise force close to zero, which suggests that the use of blown lift during landingis practical.

13. Subcellular electrical stimulation of neurons enhances the myelination of axons by oligodendrocytes

Hae Ung Lee, Agata Blasiak, Devansh Agrawal, Daniel Teh Boon Loong, Nitish V. Thakor, Angelo H. All, John S. Ho, In Hong Yang
PLOS One
2017
pdf / link / / /

@article{10.1371/journal.pone.0179642,
  author = {Lee, Hae Ung AND Blasiak, Agata AND Agrawal, Devansh R. AND Loong, Daniel Teh Boon AND Thakor, Nitish V. AND All, Angelo H. AND Ho, John S. AND Yang, In Hong},
  journal = {PLOS ONE},
  publisher = {Public Library of Science},
  title = {Subcellular electrical stimulation of neurons enhances the myelination of axons by oligodendrocytes},
  year = {2017},
  month = {07},
  volume = {12},
  url = {https://doi.org/10.1371/journal.pone.0179642},
  pages = {1-17},
  number = {7},
  doi = {10.1371/journal.pone.0179642}
}

Myelin formation has been identified as a modulator of neural plasticity. New tools are required to investigate the mechanisms by which environmental inputs and neural activity regulate myelination patterns. In this study, we demonstrate a microfluidic compartmentalized culture system with integrated electrical stimulation capabilities that can induce neural activity by whole cell and focal stimulation. A set of electric field simulations was performed to confirm spatial restriction of the electrical input in the compartmentalized culture system. We further demonstrate that electrode localization is a key consideration for generating uniform the stimulation of neuron and oligodendrocytes within the compartments. Using three configurations of the electrodes we tested the effects of subcellular activation of neural activity on distal axon myelination with oligodendrocytes. We further investigated if oligodendrocytes have to be exposed to the electrical field to induce axon myelination. An isolated stimulation of cell bodies and proximal axons had the same effect as an isolated stimulation of distal axons co-cultured with oligodendrocytes, and the two modes had a non-different result than whole cell stimulation. Our platform enabled the demonstration that electrical stimulation enhances oligodendrocyte maturation and myelin formation independent of the input localization and oligodendrocyte exposure to the electrical field.

14. Conformal phased surfaces for wireless powering of bioelectronic microdevices

Devansh Agrawal, Yuji Tanabe, Desen Weng, Andrew Ma, Stephanie Hsu, Song-Yan Liao, Zhe Zhen, Zi-Yi Zhu, Chuanbowen Sun, Zhenya Dong, Fengyuan Yang, Hung Fat Tse, Ada S. Y. Poon, John S. Ho
Nature Biomed. Eng.
2017
pdf / link / / /

@article{Agrawal2017,
  author = {Agrawal, Devansh R. and Tanabe, Yuji and Weng, Desen and Ma, Andrew and Hsu, Stephanie and Liao, Song Yan and Zhen, Zhe and Zhu, Zi Yi and Sun, Chuanbowen and Dong, Zhenya and Yang, Fengyuan and Tse, Hung Fat and Poon, Ada S.Y. and Ho, John S.},
  doi = {10.1038/s41551-017-0043},
  issn = {2157846X},
  journal = {Nature Biomedical Engineering},
  number = {3},
  pages = {1--9},
  publisher = {Macmillan Publishers Limited, part of Springer Nature.},
  title = {{Conformal phased surfaces for wireless powering of bioelectronic microdevices}},
  url = {http://dx.doi.org/10.1038/s41551-017-0043},
  volume = {1},
  year = {2017}
}

Wireless powering could enable the long-term operation of advanced bioelectronic devices within the human body. Although both enhanced powering depth and device miniaturization can be achieved by shaping the field pattern within the body, existing electromagnetic structures do not provide the spatial phase control required to synthesize such patterns. Here, we describe the design and operation of conformal electromagnetic structures, termed phased surfaces, that interface with non-planar body surfaces and optimally modulate the phase response to enhance the performance of wireless powering. We demonstrate that the phased surfaces can wirelessly transfer energy across anatomically heterogeneous tissues in large animal models, powering miniaturized semiconductor devices (<12 mm3) deep within the body (>4 cm). As an illustration of in vivo operation, we wirelessly regulated cardiac rhythm by powering miniaturized stimulators at multiple endocardial sites in a porcine animal model.

Patents

Methods and Systems for Remote Operation of Vehicles

(patent pending) The present invention provides methods and systems for remote operation of a vehicle with the capability to deal with communications jitter and intermittency. In particular, the methods and systems herein may safely predict a remote operator's intent (e.g., remote pilot) over long time scales, and up to the lost link timeout TLL.
More Info >>

Design and source code modified from Jon Barron's website. Edit here.