5632 research outputs found
Sort by
Interactive Multi-Modal Motion Planning With Branch Model Predictive Control
No full textMotion planning for autonomous robots and vehicles in presence of uncontrolled agents remains a challenging problem as the reactive behaviors of the uncontrolled agents must be considered. Since the uncontrolled agents usually demonstrate multimodal reactive behavior, the motion planner needs to solve a continuous motion planning problem under these behaviors, which contains a discrete element. We propose a branch Model Predictive Control (MPC) framework that plans over feedback policies to leverage the reactive behavior of the uncontrolled agent. In particular, a scenario tree is constructed from a finite set of policies of the uncontrolled agent, and the branch MPC solves for a feedback policy in the form of a trajectory tree, which shares the same topology as the scenario tree. Moreover, coherent risk measures such as the Conditional Value at Risk (CVaR) are used as a tuning knob to adjust the tradeoff between performance and robustness. The proposed branch MPC framework is tested on an autonomous vehicle planning problem in simulation, and on an autonomous quadruped robot alongside an uncontrolled quadruped in experiments. The result demonstrates interesting human-like behaviors, achieving a balance between safety and performance
MLNav: Learning to Safely Navigate on Martian Terrains
No full textWe present MLNav, a learning-enhanced path planning framework for safety-critical and resource-limited systems operating in complex environments, such as rovers navigating on Mars. MLNav makes judicious use of machine learning to enhance the efficiency of path planning while fully respecting safety constraints. In particular, the dominant computational cost in such safety-critical settings is running a model-based safety checker on the proposed paths. Our learned search heuristic can simultaneously predict the feasibility for all path options in a single run, and the model-based safety checker is only invoked on the top-scoring paths. We validate in high-fidelity simulations using both real Martian terrain data collected by the Perseverance rover, as well as a suite of challenging synthetic terrains. Our experiments show that: (i) compared to the baseline ENav path planner on board the Perserverance rover, MLNav can provide a significant improvement in multiple key metrics, such as a 10x reduction in collision checks when navigating real Martian terrains, despite being trained with synthetic terrains; and (ii) MLNav can successfully navigate highly challenging terrains where the baseline ENav fails to find a feasible path before timing out
Scientific multi-agent reinforcement learning for wall-models of turbulent flows
No full textThe predictive capabilities of turbulent flow simulations, critical for aerodynamic design and weather prediction, hinge on the choice of turbulence models. The abundance of data from experiments and simulations and the advent of machine learning have provided a boost to turbulence modeling efforts. However, simulations of turbulent flows remain hindered by the inability of heuristics and supervised learning to model the near-wall dynamics. We address this challenge by introducing scientific multi-agent reinforcement learning (SciMARL) for the discovery of wall models for large-eddy simulations (LES). In SciMARL, discretization points act also as cooperating agents that learn to supply the LES closure model. The agents self-learn using limited data and generalize to extreme Reynolds numbers and previously unseen geometries. The present simulations reduce by several orders of magnitude the computational cost over fully-resolved simulations while reproducing key flow quantities. We believe that SciMARL creates unprecedented capabilities for the simulation of turbulent flows
Formation of gold-bearing listvenite in the mantle section of the Neoproterozoic Bir Umq ophiolite, Western Arabian Shield, Saudi Arabia
No full textSerpentinized peridotite and associated listvenites of Neoproterozoic age outcrop in the Bir Umq area of western Saudi Arabia. The mantle section of the Bir Umq ophiolite is extensively serpentinized. Serpentinite host samples are low in Al₂O₃ (0.48–0.75 wt%) and CaO (0.28–1.24 wt%) and have high Mg# (0.90–0.92), indicating a strongly depleted mantle protolith, typically associated with supra-subduction zone environments and more specifically with fore-arc settings. Listvenite bodies of various shapes and sizes developed by alteration of serpentinite. Listvenites occupy the hanging walls of a stack of thrust faults, while serpentinite dominates the footwalls. Based on mineralogical composition and whole-rock geochemistry, the listvenites of Bir Umq are distinguished into carbonate listvenite and silica-carbonate listvenite; the latter is further divided into mineralized and non-mineralized samples. Carbonate listvenite is high in MgO, Fe₂O₃, and CaO, but depleted in SiO₂. Silica-carbonate listvenite is characterized by the presence of rhythmic banding of quartz and carbonate minerals and by the presence of fuchsite. The Bir Umq listvenites preserve various stages of the progressive alteration and metasomatic transformation of their ultramafic protoliths due to interaction with hydrothermal fluids enriched in CO₂, SiO₂, Au, K, and other fluid-mobile elements. The association with thrusting suggests that faults acted as conduits for fluids derived from metamorphism of the underlying units during subduction and obduction. Schistosity and deformation fabrics in carbonate listvenite imply that initial listvenitization took place at conditions similar to the conditions of serpentinization. On the other hand, the absence of deformation fabrics in silica-carbonate listvenite suggests that it postdates serpentinization and therefore represents a separate and later fluid infiltration event. Finally, the mineralized silica-carbonate listvenite is highly enriched in fluid-mobile elements Zn, Pb, Cu, Ag, and most notably Au; this enrichment is not correlated with silica content. This suggests that yet a third fluid infiltration event is responsible for the mineralization and Au enrichment. Au concentrations are 0.84–2.31 ng g⁻¹ in host serpentinite, 26–403 ng g⁻¹ in carbonate listvenite, 152–545 ng g⁻¹ in non-mineralized silica-carbonate listvenite, and 2286–3712 ng g⁻¹ in mineralized silica-carbonate listvenite
Inclusive and differential cross section measurements of single top quark production in association with a Z boson in proton-proton collisions at √s = 13 TeV
No full textInclusive and differential cross sections of single top quark production in association with a Z boson are measured in proton-proton collisions at a center-of-mass energy of 13 TeV with a data sample corresponding to an integrated luminosity of 138 fb⁻¹ recorded by the CMS experiment. Events are selected based on the presence of three leptons, electrons or muons, associated with leptonic Z boson and top quark decays. The measurement yields an inclusive cross section of 87.9^(+7.5)_(−7.3)(stat)^(+7.3)_(−6.0)(syst) fb for a dilepton invariant mass greater than 30 GeV, in agreement with standard model (SM) calculations and represents the most precise determination to date. The ratio between the cross sections for the top quark and the top antiquark production in association with a Z boson is measured as 2.37^(+0.56)_(−0.42)(stat)^(+0.27)_(−0.13)(syst). Differential measurements at parton and particle levels are performed for the first time. Several kinematic observables are considered to study the modeling of the process. Results are compared to theoretical predictions with different assumptions on the source of the initial-state b quark and found to be in agreement, within the uncertainties. Additionally, the spin asymmetry, which is sensitive to the top quark polarization, is determined from the differential distribution of the polarization angle at parton level to be 0.54 ± 0.16 (stat) ± 0.06 (syst), in agreement with SM predictions
A Neural Network-Assisted Euler Integrator for Stiff Kinetics in Atmospheric Chemistry
No full textAtmospheric chemistry, characterized by highly coupled sets of ordinary differential equations (ODEs), is dynamically stiff owing to the fact that both fast and slow processes exist simultaneously. We develop here a neural network-assisted Euler integrator for the kinetics of atmospheric chemical reactions. We show that the integral kernel of the chemical reaction system can be represented by a neural network. The stiff kinetics of the atmospheric H₂O₂/OH/HO₂ system, involving 3 species and 4 reactions, and a simplified air pollution mechanism, involving 20 species and 25 reactions, are developed here in detail as illustrations of the neural network Euler integrator. The algorithm developed accelerates the numerical integration of large sets of coupled stiff ODEs by at least one order of magnitude by avoiding the intensive linear algebra that is required in traditional stiff ODE solvers; moreover, the mechanism-specific neural network-assisted algorithm can be readily coupled to other modules in a three-dimensional atmospheric chemical transport model
Do Major-Power Interventions Encourage the Onset of Civil Conflict? A Structural Analysis
No full textWhat is the impact of major-power intervention on civil-war onset? A considerable hurdle to answering this question is gauging expectations about future intervention on the eve of conflict. We tackle this challenge by developing a unified theoretical and empirical model of civil-war onset and subsequent intervention. Our model allows for strategic interdependence among interveners, and our empirical strategy enables estimation of intervention expectations from equilibrium behavior. We fit the model to civil war and intervention data from the second half of the twentieth century and find that major-power intervention is primarily characterized by strategic complementarities—for example, cost sharing among allies or competition for control among rivals—rather than free-riding incentives. Through counterfactual experiments, we show that commitments to decreased intervention would raise the risk of civil war worldwide, whereas increased intervention would have little effect. Our results suggest that coordination among major powers is maximally deterring civil conflict
Space-time controlled metasurfaces for active multi-channel beam shaping
No full textMetasurfaces form a powerful approach to realizing compact and lightweight optical elements, which can be integrated into smart glasses and head-mounted displays. In this talk, we explore the opportunities that arise with electronically programmable active metasurfaces, which are simultaneously modulated in both space and time. Using electro-optical effects, our group has previously demonstrated metasurfaces that control the spatial features of light. By doing so, we have been able to realize multifunctional optical elements that can achieve beam focusing and steering with a high signal-to-noise ratio.1,2 However, in this quasi-static operation regime, the applied signal is not varied in time. The introduction of time modulation additionally allows the creation of higher-order frequency harmonics that provide control over the spectral content of the scattered light. We implement time-modulated metasurfaces by integrating an indium tin oxide (ITO) based, electro-optically tunable metasurface operating at 1550 nm into a radiofrequency network. Each metasurface element is modulated at up to 100 MHz to generate frequency harmonics that are well separated from the central frequency. With the use of additional nonresonant phase shifters, we engineer space-time modulated wavefronts that allow us to control a four-dimensional design space. Finally, we demonstrate a metasurface architecture consisting of interdigitated subarrays that are independently controlled using distinct spatiotemporal phase fronts. With this, we are able to demonstrate simultaneous and independent shaping of beams at distinct frequencies using a single chip. We foresee that this technology will have direct implications on the future of multi-channel optical communication networks used in AR/VR systems
Periodic spatial patterning with a single morphogen
No full textMulticellular development employs periodic spatial patterning to generate repetitive structures such as digits, vertebrae, and teeth. Turing patterning has long provided a key paradigm for understanding such systems. The simplest Turing systems are believed to require at least two signals, or morphogens, that diffuse and react to spontaneously generate periodic patterns. Here, using mathematical modeling, we show that a minimal circuit comprising an intracellular positive feedback loop and a single diffusible morphogen is sufficient to generate stable, long-range spatially periodic cellular patterns. The model considers cells as discrete entities as a key feature, and incorporates transient boundary conditions. Linear stability analysis reveals that this single-morphogen Turing circuit can support a broad range of spatial wavelengths, including fine-grain patterns similar to those generated by classic lateral inhibition systems. Further, signals emanating from a boundary can initiate and stabilize propagating modes with a well-defined spatial wavelength. Once formed, patterns are self-sustaining and robust to noise. Finally, while noise can disrupt patterning in pre-patterned regions, its disruptive effect can be overcome by a bistable intracellular circuit loop, or by considering patterning in the context of growing tissue. Together, these results show that a single morphogen can be sufficient for robust spatial pattern formation, and should provide a foundation for engineering pattern formation in the emerging field of synthetic developmental biology
Spatially resolved gas-phase metallicity in FIRE-2 dwarfs: late-time evolution of metallicity relations in simulations with feedback and mergers
No full textWe present an analysis of spatially resolved gas-phase metallicity relations in five dwarf galaxies (Mₕₐₗₒ ≈ 10¹¹ M_⊙, M⋆ ≈ 10^(8.8)−10^(9.6) M_⊙) from the FIRE-2 (Feedback in Realistic Environments) cosmological zoom-in simulation suite, which include an explicit model for sub-grid turbulent mixing of metals in gas, near z ≈ 0, over a period of 1.4 Gyr, and compare our findings with observations. While these dwarf galaxies represent a diverse sample, we find that all simulated galaxies match the observed mass–metallicity (MZR) and mass–metallicity gradient (MZGR) relations. We note that in all five galaxies, the metallicities are effectively identical between phases of the interstellar medium (ISM), with 95 per cent of the gas being within ±0.1 dex between the cold and dense gas (T 1 cm⁻³), ionized gas (near the HαT ≈ 10⁴ K ridge-line), and nebular regions (ionized gas where the 10 Myr-averaged star formation rate is non-zero). We find that most of the scatter in relative metallicity between cold dense gas and ionized gas/nebular regions can be attributed to either local starburst events or metal-poor inflows. We also note the presence of a major merger in one of our galaxies, m11e, with a substantial impact on the metallicity distribution in the spatially resolved map, showing two strong metallicity peaks and triggering a starburst in the main galaxy