Search for: All records

Fallen logs acting as a seedbed for trees to aid the regeneration of vegetation is a common ecological strategy in modern forests. However, the origin, occurrence, and evolution of this nurse log strategy in the geological time is unclear. Here we report a ca. 310-millionyear-old permineralized cordaitalean tree trunk from the Moscovian (Pennsylvanian, upper Carboniferous) Benxi Formation in Yangquan City, Shanxi Province, North China, with evidence of probable cordaitalean rootlets growing inside the trunk. The specimen is interpreted as a nurse log for regeneration of cordaitaleans in coastal lowlands. It provides the first glimpse of plant-plant facilitative interaction between Pennsylvanian cordaitaleans in Cathaysia. We interpret that the Moscovian cordaitalean seedlings preferentially established on the fallen log owing to the ability of the rotting wood to store fresh water. The nurse log provided a stable substrate in an environment with episodic salinity and/or water table variations. In combination with previous records, it is suggested that a sophisticated terrestrial ecosystem with multiple interactions between plants and other organisms have developed on the central North China Craton no later than the Middle Pennsylvanian. 

Tensegrity robots, composed of rigid rods and flexible cables, exhibit high strength-to-weight ratios and significant deformations, which enable them to navigate unstructured terrains and survive harsh impacts. They are hard to control, however, due to high dimensionality, complex dynamics, and a coupled architecture. Physics-based simulation is a promising avenue for developing locomotion policies that can be transferred to real robots. Nevertheless, modeling tensegrity robots is a complex task due to a substantial sim2real gap. To address this issue, this paper describes a Real2Sim2Real (R2S2R) strategy for tensegrity robots. This strategy is based on a differentiable physics engine that can be trained given limited data from a real robot. These data include offline measurements of physical properties, such as mass and geometry for various robot components, and the observation of a trajectory using a random control policy. With the data from the real robot, the engine can be iteratively refined and used to discover locomotion policies that are directly transferable to the real robot. Beyond the R2S2R pipeline, key contributions of this work include computing non-zero gradients at contact points, a loss function for matching tensegrity locomotion gaits, and a trajectory segmentation technique that avoids conflicts in gradient evaluation during training. Multiple iterations of the R2S2R process are demonstrated and evaluated on a real 3-bar tensegrity robot.