Search for: All records

In this paper, we introduce a model of evolution and learning in robots that co-optimizes a distribution of latent design vectors (genotypes) and a mixture of control experts (neural modules), which are gated by the latent coordinates of each decoded design (phenotype). This provides a scalable alternative to co-design algorithms that either train an individual policy for every robot, which is inefficient, or a monolithic universal controller for all robots, which results in overly conservative structures and behaviors. Our approach lies somewhere between these two extremes, preserving ancestral knowledge in a unified yet modular framework in which different body plans activate and deactivate different combinations of learned sensorimotor circuits for goal-directed behavior. This allows one part of the controller to be overhauled to better suit new species of designs as they emerge without disrupting the hard-earned knowledge contained within other expert modules. It also allows pretrained expert policies to be directly plugged into the mixture, which can steer evolution into otherwise unexplored areas of latent space containing desired morphological traits. We refer to this process as "evo by demo" and explore how it may be used to guide freeform evolution toward canonical structures defined by the pretrained model. Videos and code can be found at: https://eco-moe.github.io. 

Aims.In this work, we derive upper limits on the strength of the near-core magnetic field in intermediate-mass stars, based on the suppression of high-orderg-modes by a critical magnetic field. We considered both purely poloidal magnetic fields and mixed geometries with the inclusion of a toroidal component. We examined how the upper limits on magnetic field strengths are affected by the degree and azimuthal order of the oscillations, as well as the magnetic field configuration. We compared our results with observations of red giant stars, which are the evolved counterparts of intermediate-mass main sequence stars, to assess the consistency of the field strengths across evolutionary stages. Methods.We considered twoγDoradus stars hosting high-orderg-modes and an evolvedδScuti star with mixed modes. These targets have prior mode identification from observations, enabling us to use these mode frequencies to probe the near-core region. We determined the best structural model from their stellar parameters through grid-based modeling with MESA. Frequencies for the best models were extracted using GYRE and matched to the observed modes. The critical magnetic fields for all calculated frequencies in our models were obtained from the Dedalus code and used to infer an upper limit on the near-core field strength. Results.We find an upper limit on the near-core radial field strength ofB_r≈ 130 kG andB_r≈ 13 kG, assuming a dipole field configuration, for the twoγDoradus stars KIC 3127996 and KIC 5876187, respectively. For 44 Tau, analysis of mixed modes yields a field strength ofB_r≈ 1771 kG. Different magnetic field configurations and mode degrees lead to different estimates. The results for the radial component of the magnetic field in the main sequenceγDoradus stars are consistent with estimates of magnetic field strengths in red giant stars that assume an internal field generated by a core dynamo, although the stronger of the two inferred magnetic fields may require some enhancement by a fossil field. The toroidal component does not affectg-modes significantly and is required to be more than 200 times stronger than the radial component to suppressg-modes. We find evidence of rotational modulation in the light curves of the twoγDoradus stars, which we use to demonstrate that the rotation rate is consistent with rigid body rotation in the envelope. 

Predicting future walking joint kinematics is crucial for assistive device control, especially in variable walking environments. Traditional optical motion capture systems provide kinematics data but require laborious post-processing, whereas IMU based systems provide direct calculations but add delays due to data collection and algorithmic processes. Predicting future kinematics helps to compensate for these delays, enabling the system real-time. Furthermore, these predicted kinematics could serve as target trajectories for assistive devices such as exoskeletal robots and lower limb prostheses. However, given the complexity of human mobility and environmental factors, this prediction remains to be challenging. To address this challenge, we propose the Dual-ED-Attention-FAM-Net, a deep learning model utilizing two encoders, two decoders, a temporal attention module, and a feature attention module. Our model outperforms the state-of-the-art LSTM model. Specifically, for Dataset A, using IMUs and a combination of IMUs and videos, RMSE values decrease from 4.45° to 4.22° and from 4.52° to 4.15°, respectively. For Dataset B, IMUs and IMUs combined with pressure insoles result in RMSE reductions from 7.09° to 6.66° and from 7.20° to 6.77°, respectively. Additionally, incorporating other modalities alongside IMUs helps improve the performance of the model.