A 2017 IEEE Journal of Solid-State Circuits paper introduced the BrainSoC, a fully integrated battery-powered system-on-chip in 40-nm CMOS intended to provide onboard closed-loop control for the Harvard RoboBee, an insect-scale flapping-wing pico-aerial vehicle, and explicitly deferred the demonstration of sustained autonomous closed-loop flight to future work. The subsequent literature shows that the objective has been approached piecemeal but not achieved: untethered flight was demonstrated in 2019 with photovoltaic power under intense external illumination and no onboard sensing or computation; a Science Robotics 2025 platform (750 mg, 1000-second hovering) still relied on offboard power and motion-capture sensing; a February 2026 arXiv work demonstrated hovering and trajectory tracking with fully onboard sensing and computation (1.29 g) but offboard power; and a February 2026 unrefereed preprint claims sustained untethered sub-gram flight with a 350 mg onboard battery but only mechanical self-stabilization, with no onboard sensing or computation autonomy claimed. The surviving open problem is the integrated one: sustained, stable, closed-loop flight of an insect-scale flapping-wing microrobot in which energy storage, state sensing, and estimation/control computation are all onboard, together with quantitative characterization of system-level closed-loop flight performance (stability, maneuverability, and endurance).

Background

A flapping-wing micro aerial vehicle (FWMAV) is a flying robot of at most a few grams that generates lift and control torques by beating wings at high frequency, the way insects do. At the insect scale — roughly one gram and below, down to the ~100 mg Harvard RoboBee — flapping flight is aerodynamically effective, but every onboard resource becomes scarce: the vehicle's actuators consume on the order of 100 mW, total payload capacity is a few hundred milligrams, and the vehicle's body dynamics are fast and intrinsically unstable, so stable hovering requires feedback control at high update rates. Because of this, laboratory FWMAVs have historically been restrained by three kinds of tethers: electrical wires delivering actuator power from offboard high-voltage amplifiers, motion-capture cameras supplying position and attitude feedback from outside the flight volume, and desktop computers running the control loops. Eliminating these dependencies is conventionally decomposed into three 'autonomies': energy autonomy (onboard energy storage and power conversion), sensing autonomy (onboard measurement of the states needed for control), and computation autonomy (onboard real-time estimation and control). The first controlled flight of an insect-scale robot (Ma, Chirarattananon, Fuller, and Wood, Science, 2013) achieved stable hovering of an ~80 mg RoboBee but used all three tethers. The source paper (Zhang, Lok, Tong, Lee, Reagen, Chaput, Duhamel, Wood, Brooks, and Wei, IEEE Journal of Solid-State Circuits, 2017) addressed the computation leg: the BrainSoC, a battery-powered system-on-chip — a single integrated circuit combining a processor, specialized hardware, and power management — fabricated in a 40-nm CMOS process (CMOS, complementary metal–oxide–semiconductor, being the standard silicon technology used to build commercial integrated circuits). It integrates a general-purpose processor, a hardware accelerator for flight control, sensing interfaces, and power management, and was designed to run the complete closed-loop control of a RoboBee-class pico-aerial vehicle within a power budget of tens of microwatts. The authors explicitly deferred the full multi-chip system assembly and the experimental demonstration of sustained autonomous closed-loop flight; they noted that leakage was significant in the chosen high-performance process with shared power domains, deferring leakage-reduction redesign to a future SoC; and they reported that no off-the-shelf battery met the vehicle's simultaneous mass and energy-density needs — the paper's summary reports a design guideline of roughly 150 mJ/mg — leaving onboard energy storage as an open constraint rather than a demonstrated capability. In the years since, the field has dismantled these tethers one at a time: laser-photovoltaic power enabled brief untethered liftoff of a 190 mg vehicle (James, Iyer, Chukewad, Gollakota, and Fuller, ICRA 2018) and then sustained untethered hovering of a 259 mg vehicle under illumination about three times sunlight intensity (Jafferis, Helbling, Karpelson, and Wood, Nature 2019), though with no onboard sensing, no onboard computation, and an effective dependency on the external light source. Avionics miniaturization continued (for example, a 78.4 mg, 15 mW sensor suite described by Yu, Tran, Li, Weber, Talwekar, and Fuller, arXiv 2025), and a 750 mg vehicle achieved 1000-second hovering with sub-centimeter trajectory accuracy (Kim, Hsiao, Ren, Huang, and Chen, Science Robotics 2025) — still with offboard power and motion-capture sensing, with the authors naming sensing and power autonomy as the open opportunities. Most recently, a 1.29 g robot demonstrated hovering and trajectory tracking using only onboard sensors and a single microcontroller core, including a 30-second flight outside any motion-capture volume with obstacle avoidance and landing on a flower under human high-level commands (Hsiao, Kieu, Guan, Kim, Cai, Matteson, How, Helbling, and Chen, arXiv 2026) — while its actuator power remained offboard, the authors estimating that adding onboard power electronics and batteries would require roughly 1–1.5 g of additional payload. In parallel, an unrefereed preprint (Xiao et al., Research Square, 2026) reports a 984 mg piezoelectric vehicle with a 350 mg onboard battery achieving untethered flights exceeding 240 seconds, stabilized mechanically rather than by onboard sensing and computation. What no work has demonstrated is the combination: an insect-scale flapping-wing vehicle whose flight is simultaneously sustained by onboard stored energy, stabilized by onboard sensing, and computed by onboard processing.

Problem Statement

Achieve and characterize sustained, stable, closed-loop flight of an insect-scale flapping-wing microrobot — a vehicle of roughly one gram total mass, in the scale class running from the ~100 mg RoboBee target of the source system-on-chip up to the 1.29 gram state-of-the-art 2026 platform — with all three autonomies onboard simultaneously: (i) energy autonomy, meaning onboard electrochemical energy storage and power conversion with no wired power, laser illumination, or other offboard energy delivery, such that flight duration is limited by stored onboard energy rather than by stability loss or operator intervention; (ii) sensing autonomy, meaning all states used for flight stabilization are measured by sensors carried on the vehicle, with no motion-capture or other external sensing in the control loop; and (iii) computation autonomy, meaning the state estimation and control computation that stabilize the vehicle execute onboard in real time, with any offboard radio link restricted to optional high-level, non-stabilizing commands. A resolving answer must deliver a concrete artifact — either an integrated system design for such a vehicle, with a quantified mass and power budget and a closed-loop stability and endurance analysis, or a specified experimental flight protocol with measured results — and must characterize that system's closed-loop flight performance: end-to-end closed-loop stability and positional accuracy, maneuverability, and flight endurance limited by stored onboard energy. It must also report the integrated system's mass and power budget, so that the trade-offs among battery mass, sensing payload, and control performance at insect scale are quantified.

References

  1. Zhang, X.; Lok, M.; Tong, T.; Lee, S. K.; Reagen, B.; Chaput, S.; Duhamel, P.-E. J.; Wood, R. J.; Brooks, D.; Wei, G.-Y. 'A Fully Integrated Battery-Powered System-on-Chip in 40-nm CMOS for Closed-Loop Control of Insect-Scale Pico-Aerial Vehicle.' IEEE Journal of Solid-State Circuits 52(9):2374–2387, 2017. DOI: 10.1109/jssc.2017.2705170. https://doi.org/10.1109/jssc.2017.2705170
  2. Zhang, X.; Lok, M.; Tong, T.; Chaput, S.; Lee, S. K.; Reagen, B.; Lee, H.; Brooks, D.; Wei, G.-Y. 'A multi-chip system optimized for insect-scale flapping-wing robots.' 2015 Symposium on VLSI Circuits (VLSI Circuits), pages C152–C153, 2015. DOI: 10.1109/vlsic.2015.7231246. https://doi.org/10.1109/vlsic.2015.7231246
  3. Ma, K. Y.; Chirarattananon, P.; Fuller, S. B.; Wood, R. J. 'Controlled Flight of a Biologically Inspired, Insect-Scale Robot.' Science 340:603–607, 2013. DOI: 10.1126/science.1231806. https://doi.org/10.1126/science.1231806
  4. James, J.; Iyer, V.; Chukewad, Y.; Gollakota, S.; Fuller, S. B. 'Liftoff of a 190 mg Laser-Powered Aerial Vehicle: The Lightest Wireless Robot to Fly.' 2018 IEEE International Conference on Robotics and Automation (ICRA), 2018. DOI: 10.1109/icra.2018.8460582. https://doi.org/10.1109/icra.2018.8460582
  5. Jafferis, N. T.; Helbling, E. F.; Karpelson, M.; Wood, R. J. 'Untethered flight of an insect-sized flapping-wing microscale aerial vehicle.' Nature 570:491–495, 2019. DOI: 10.1038/s41586-019-1322-0. https://doi.org/10.1038/s41586-019-1322-0
  6. Yu, Z.; Tran, J.; Li, C.; Weber, A.; Talwekar, Y. P.; Fuller, S. 'TinySense: A Lighter Weight and More Power-efficient Avionics System for Flying Insect-scale Robots.' arXiv:2501.03416, 2025. https://arxiv.org/abs/2501.03416
  7. Kim, S.; Hsiao, Y.-H.; Ren, Z.; Huang, J.; Chen, Y. 'Acrobatics at the insect scale: A durable, precise, and agile micro-aerial robot.' Science Robotics 10(98), 2025. DOI: 10.1126/scirobotics.adp4256. https://doi.org/10.1126/scirobotics.adp4256
  8. Muller, H.; Palossi, D.; Mach, S.; Conti, F.; Benini, L. 'Fünfiiber-Drone: A Modular Open-Platform 18-grams Autonomous Nano-Drone.' 2021 Design, Automation & Test in Europe Conference & Exhibition (DATE), pages 1610–1615, 2021. DOI: 10.23919/date51398.2021.9474262. https://doi.org/10.23919/date51398.2021.9474262
  9. Hsiao, Y.-H.; Kieu, Q. P.; Guan, Z.; Kim, S.; Cai, J.; Matteson, O.; How, J. P.; Helbling, E. F.; Chen, Y. 'Controlled Flight of an Insect-Scale Flapping-Wing Robot via Integrated Onboard Sensing and Computation.' arXiv:2602.08328, 2026. https://arxiv.org/abs/2602.08328
  10. Xiao, D.; Lu, X.; Chen, J.; Wu, Y.; Chen, Y.; Yin, C.; Deng, Z.; Qu, C.; Lu, K.; Wu, X. 'An untethered sub-gram flapping-wing flight system achieves sustained self-stabilizing and controllable flight.' Research Square preprint (in review), 2026. DOI: 10.21203/rs.3.rs-8253091/v1. https://doi.org/10.21203/rs.3.rs-8253091/v1