Post-disaster mountain UAV search-and-rescue path planning must balance range, energy, threats, search reward, and attitude jitter, forming a multi-constraint, multi-objective problem in 3D terrain with disconnected feasible regions. Existing artificial lemming algorithm (ALA) variants are single-objective, suffering from solution loss, fixed perturbations, and limited diversity. We propose MO-ALA, a multi-objective ALA with dual-layer historical memory and adaptive t-distribution perturbation. Memory improves convergence, while the t-distribution couples a time-varying tail-shape parameter with a decaying amplitude envelope, enabling broad early exploration and precise late search; archive pruning yields a weight-adjustable approximate Pareto set. On ZDT1–ZDT3, MO-ALA outperforms five baselines in IGD; against its ablation variants, however, it is significantly worse than Variant A on ZDT1 and nominally better on DTLZ3 (p = 0.026, not surviving the Holm correction), with no significant difference on the other four benchmarks. The contribution is scoped to non-deceptive fronts: on deceptive DTLZ1/DTLZ3, MO-ALA (like all tested ALA variants) fails to converge (HV ≈ 0). A crossover-based hybrid restores convergence but degrades ZDT precision, remaining an optional remedy. Terrain-scenario ablation attributes the detected benefit to the memory layer, while the t-distribution adds no measurable cost or gain. Path length increases of ~ 21%/6% yield ~ 43%/40% reductions in safety and search-reward cost, respectively, supporting post-disaster deployment.
Finite-resolution phase-only antenna synthesis is a constrained, multimodal combinatorial search problem in which small changes in digital phase codes can substantially alter interference suppression. This study applies the Enhanced Rüppell’s Fox Optimizer (eRFO) to null steering in a 20-element linear array of half-wave dipoles with fixed half-wavelength spacing and unity excitation amplitudes. The bioinspired search framework combines exploration and exploitation with adaptive control, chaotic diversity, opposition-based learning, and local refinement; candidate evaluation is restricted to feasible digital phase states. Conjugate-symmetric excitation reduces the number of independent optimization variables from 20 to 10. The radiation model combines the array factor with the isolated half-wave dipole element factor. Numerical examples examine sidelobe control, single and multiple point nulls, 4-bit and 8-bit phase resolution, and angular sector-null suppression. The reported 8-bit single-null case reaches −93.48 dB, while the dual- and triple-null cases place all prescribed nulls below −91 dB and −92 dB, respectively. The 4-bit single-null example reaches −73.64 dB. For the angular sector-null case from 11.75° to 16.75°, eRFO attains a maximum sampled level of −70.41 dB and is compared with genetic algorithm (GA), differential evolution (DE), and particle swarm optimization (PSO) solutions. Statistics over 50 independent runs, a mechanism ablation, a deterministic benchmark with a posteriori quantization, and sensitivity analyses under random excitation errors and mutual coupling complement these idealized single-frequency results. The contribution is the adaptation of bioinspired search to a discrete electromagnetic design problem, with explicit trade-offs among null suppression, sidelobe control, and phase resolution.
High-speed human running data are difficult to acquire, limiting the availability of motion priors for high-speed humanoid running. To reduce dependence on motion capture and retargeting, this study proposes a framework that combines trajectory optimization with reinforcement learning. A planar multibody model generates reference motions across speeds subject to contact, periodicity, and actuator constraints. The optimized motions guide the training of a three-dimensional running policy through adversarial imitation and proximal policy optimization. Hardware experiments demonstrate treadmill running by the full-size BOLT humanoid at 10m/s. In 2s simulation windows under a 10m/s forward velocity command, comparison with a policy trained on retargeted human motions shows a more balanced distribution of joint loads relative to actuator capacity. The coefficient of variation of joint root-mean-square (RMS) utilization decreases from 0.397 to 0.317 across twelve joints, a reduction of 20.1%, with a corresponding reduction of 35.6% across the six leg joints. These results support trajectory optimization as an effective source of motion priors for high-speed humanoid running and demonstrate a route from reference generation to hardware deployment without high-speed human motion data.
Scyphozoan jellyfish exhibit the highest locomotive efficiency in the animal kingdom, making them of particular interest in fluid dynamics and bioinspired robotics. Despite this, prevalent analytical models of jellyfish swimming have been based on the swimming traits of hydrozoan jellyfish, which utilize jet propulsion, rather than scyphozoan jellyfish, which utilize paddling propulsive methods. Additionally, while stroke frequency is a driving variable in speeds achieved by undulatory swimmers, a similar dependence has not been thoroughly demonstrated for jellyfish. This work investigates the relationship between stroke frequency and swimming speeds in two species of scyphozoan jellyfish, Aurelia aurita and Cassiopea xamachana. An experimental study was conducted using a biohybrid technique that controls the muscle contraction frequency of freely swimming, live jellyfish with portable, implanted microelectronics. Jellyfish were stimulated to swim at nine frequencies from 0.3 to 0.8 Hz, and swimming speeds were measured from video recordings in a 2.4 m tall water tank. It was found that despite differences in their natural swimming frequencies, the Aurelia and Cassiopea displayed similar speed-frequency relationships with peak swimming speeds occurring at 0.559±0.003 Hz and 0.501±0.006 Hz, respectively. The difference in natural stroke frequency displayed by scyphomedusae, despite the shared relationship between swimming speed and stroke frequency in these two species, suggests that natural stroke frequency may be more related to other functions such as filter feeding, or respiration, rather than locomotion. A new analytical model developed for scyphozoan, paddling jellyfish was shown to have closer agreement with the experimental results than existing models based on jet propulsion (peak-normalized root mean square errors of 24.8% for the Aurelia and 24.2% for the Cassiopea). The model demonstrated the driving factors in the relationship between swimming speed and stroke frequency to be the speed of the jellyfish bell margin and changes in body kinematics with stroke frequency. This further-defined relationship between swimming frequency and speed in scyphozoan jellyfish has implications for speed control and modeling of both biohybrid and bioinspired robotics.
Sky-region recognition using a bio-inspired polarization sensor helps reduce occlusion-induced errors in polarization orientation and supports applications such as Global Navigation Satellite System (GNSS) non-line-of-sight error mitigation and atmospheric irradiance estimation. However, retaining usable sky measurements while rejecting non-sky interference remains challenging in scenes with extensive occlusions. This paper proposes a sky-region recognition and polarization orientation method that combines multi-scale grayscale gradients with connected-component extraction. Multi-scale gradients are computed from the grayscale image derived from the polarization measurements. Gradient, intensity, and degree of polarization (DOP) thresholds are then combined to obtain an initial binary sky mask. Connected-component analysis is then performed, and the largest component is retained as the final mask, which is used to select the polarization measurements for heading estimation. Experimental results on the unmanned platform in scenes containing building, vegetation, and pedestrian occlusions show that the proposed algorithm achieves a heading root-mean-square error (RMSE) of 0.48°, corresponding to a 36.0% reduction relative to the DOP-gradient-based method.
