Q-Learning to navigate turbulence without a map DOI Open Access
Marco Rando,

Martin James,

Alessandro Verri

et al.

Published: Dec. 16, 2024

We consider the problem of olfactory searches in a turbulent environment. focus on agents that respond solely to odor stimuli, with no access spatial perception nor prior information about odor. ask whether navigation target can be learned robustly within sequential decision making framework. develop reinforcement learning algorithm using small set interpretable states and train it realistic cues. By introducing temporal memory, we demonstrate two salient features traces, discretized few states, are sufficient learn plume. Performance is dictated by sparse nature odors. An optimal memory exists which ignores blanks plume activates recovery strategy outside obtain best performance letting their show mostly casting cross wind, similar behavior observed flying insects. The robust substantial changes plumes, suggesting minor parameter tuning may adapt different environments.

Language: Английский

Q-learning with temporal memory to navigate turbulence DOI Open Access
Marco Rando,

Martin James,

Alessandro Verri

et al.

Published: March 5, 2025

We consider the problem of olfactory searches in a turbulent environment. focus on agents that respond solely to odor stimuli, with no access spatial perception nor prior information about odor. ask whether navigation target can be learned robustly within sequential decision making framework. develop reinforcement learning algorithm using small set interpretable states and train it realistic cues. By introducing temporal memory, we demonstrate two salient features traces, discretized few states, are sufficient learn plume. Performance is dictated by sparse nature odors. An optimal memory exists which ignores blanks plume activates recovery strategy outside obtain best performance letting their show mostly casting cross wind, similar behavior observed flying insects. The robust substantial changes plumes, suggesting minor parameter tuning may adapt different environments.

Language: Английский

Citations

1
The fruit fly, Drosophila melanogaster , as a microrobotics platform DOI Creative Commons

Kenichi Iwasaki,

Charles Neuhauser,

Chris Stokes

et al.

Proceedings of the National Academy of Sciences, Journal Year: 2025, Volume and Issue: 122(15)

Published: April 8, 2025

Engineering small autonomous agents capable of operating in the microscale environment remains a key challenge, with current systems still evolving. Our study explores fruit fly, Drosophila melanogaster , classic model system biology and species adept at interaction, as biological platform for microrobotics. Initially, we focus on remotely directing walking paths flies an experimental arena. We accomplish this through two distinct approaches: harnessing flies’ optomotor response optogenetic modulation its olfactory system. These techniques facilitate reliable repeated guidance between arbitrary spatial locations. guide along predetermined trajectories, enabling them to scribe patterns resembling textual characters their locomotion. enhance olfactory-guided navigation additional activation attraction-inducing mushroom body output neurons. extend control collective behaviors shared spaces constrained maze-like environments. further use our technique enable carry load across designated points space, establishing upper bound weight-carrying capabilities. Additionally, demonstrate that visual can novel interactions objects, showing consistently relocate spherical object over significant distances. Last, multiagent formation control, alternating patterns. Beyond expanding tools available microrobotics, these behavioral contexts provide insights into neurological basis behavior flies.

Language: Английский

Citations

0
Optimal trajectories for Bayesian olfactory search in turbulent flows: The low information limit and beyond DOI
R. A. Heinonen, Luca Biferale, Antonio Celani

et al.

Physical Review Fluids, Journal Year: 2025, Volume and Issue: 10(4)

Published: April 9, 2025

Language: Английский

Citations

0
Inhibitory control explains locomotor statistics in walking Drosophila DOI Creative Commons
Hannah Gattuso,

Karin A. van Hassel,

Jacob D. Freed

et al.

Proceedings of the National Academy of Sciences, Journal Year: 2025, Volume and Issue: 122(16)

Published: April 17, 2025

In order to forage for food, many animals regulate not only specific limb movements but the statistics of locomotor behavior, switching between long-range dispersal and local search depending on resource availability. How premotor circuits is clear. Here, we analyze model their modulation by attractive food odor in walking Drosophila . Food evokes three motor regimes flies: baseline walking, upwind running during odor, behavior following loss. During search, find that flies adopt higher angular velocities slower ground speeds turn longer periods same direction. We further different mean speed these state changes influence length odor-evoked runs. next developed a simple neural control suggests contralateral inhibition plays key role regulating statistical features locomotion. As fly connectome predicts decussating inhibitory neurons lateral accessory lobe (LAL), gained genetic access subset tested effects behavior. identified one population whose activation induces all signature regulates velocity at offset. second population, including single LAL neuron pair, bidirectionally speed. Together, our work develops biologically plausible computational architecture captures locomotion across behavioral states identifies substrates computations.

Language: Английский

Citations

0
Inhibitory control of locomotor statistics in walkingDrosophila DOI Creative Commons
Hannah Gattuso,

Karin A. van Hassel,

Jacob D. Freed

et al.

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: April 16, 2024

In order to forage for food, many animals regulate not only specific limb movements but the statistics of locomotor behavior over time, example switching between long-range dispersal behaviors and more localized search depending on availability resources. How pre-motor circuits such is clear. Here we took advantage robust changes in evoked by attractive odors walking

Language: Английский

Citations

3
Olfactory cues and memories in animal navigation DOI
Thierry Emonet, Massimo Vergassola

Nature Reviews Physics, Journal Year: 2024, Volume and Issue: 6(4), P. 215 - 216

Published: March 8, 2024

Language: Английский

Citations

2
Insect Flight: State of the Field and Future Directions DOI
Lisa A. Treidel, Kevin D. Deem, Mary K. Salcedo

et al.

Integrative and Comparative Biology, Journal Year: 2024, Volume and Issue: 64(2), P. 533 - 555

Published: July 8, 2024

The evolution of flight in an early winged insect ancestral lineage is recognized as a key adaptation explaining the unparalleled success and diversification insects. Subsequent transitions modifications to machinery, including secondary reductions losses, also play central role shaping impacts insects on broadscale geographic ecological processes patterns present future. Given importance flight, there has been centuries-long history research debate evolutionary origins biological mechanisms flight. Here, we revisit this from interdisciplinary perspective, discussing recent discoveries regarding developmental origins, physiology, biomechanics, neurobiology sensory control diverse set models. We identify major outstanding questions yet be addressed provide recommendations for overcoming current methodological challenges faced when studying which will allow field continue move forward new exciting directions. By integrating mechanistic work into contexts, hope that synthesis promotes stimulates efforts necessary close many existing gaps about causes consequences evolution.

Language: Английский

Citations

2
Bifurcation enhances temporal information encoding in the olfactory periphery DOI Open Access
Kiri Choi,

Will Rosenbluth,

Isabella R. Graf

et al.

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: May 31, 2024

Living systems continually respond to signals from the surrounding environment. Survival requires that their responses adapt quickly and robustly changes in One particularly challenging example is olfactory navigation turbulent plumes, where animals experience highly intermittent odor while concentration varies over many length- timescales. Here, we show theoretically

Language: Английский

Citations

1
Discovering and exploiting active sensing motifs for estimation with empirical observability DOI Creative Commons
Benjamin Cellini, Burak Boyacıoğlu, S. David Stupski

et al.

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: Nov. 6, 2024

ABSTRACT Organisms and machines must use measured sensory cues to estimate unknown information about themselves or their environment. Cleverly applied sensor motion can be exploited enrich the quality of data improve estimation. However, a major barrier modeling such active sensing problems is lack empirical, yet rigorous, tools for quantifying relationship between movement estimation performance. Here, we introduce “BOUNDS: Bounding Observability Uncertain Nonlinear Dynamic Systems”. BOUNDS discover patterns that increase reduce uncertainty in either real simulated data. Crucially, it suitable high dimensional partially observable nonlinear systems with noise. We demonstrate through case study on how flying insects wind properties, showing specific motifs Additionally, present framework refine sporadic estimates from sensing. When combined an artificial neural network, show gained via Drosophila flight trajectories precise direction Collectively, our work will help decode organisms inform design algorithms machines.

Language: Английский

Citations

1
Bifurcation Enhances Temporal Information Encoding in the Olfactory Periphery DOI Creative Commons
Kiri Choi,

Will Rosenbluth,

Isabella R. Graf

et al.

PRX Life, Journal Year: 2024, Volume and Issue: 2(4)

Published: Nov. 12, 2024

Living systems continually respond to signals from the surrounding environment. Survival requires that their responses adapt quickly and robustly changes in One particularly challenging example is olfactory navigation turbulent plumes, where animals experience highly intermittent odor while concentration varies over many length- timescales. Here, we show theoretically receptor neurons (ORNs) can exploit proximity a bifurcation point of firing dynamics reliably extract information about timing intensity fluctuations signal, which have been shown be critical for odor-guided navigation. Close bifurcation, system intrinsically invariant signal variance, timing, duration, transferred efficiently. Importantly, find maintained by mean adaptation alone therefore does not require any additional feedback mechanism or fine-tuning. Using biophysical model with calcium-based feedback, demonstrate this explain measured characteristics ORNs. Published American Physical Society 2024

Language: Английский

Citations

1