Soil models are widely used to characterize the reflectance properties of dry and wet soils. By considering detailed physical processes, the improved multilayer radiative transfer model of soil reflectance (MARMIT-2) model significantly improves the accuracy of simulating wet soil properties. However, the MARMIT-2 model relies on measured dry soil reflectance as an input, which limits its applicability in practical scenarios, especially when detailed information about specific soils is unavailable. To address this issue, this study first evaluated the ability of the general spectral vector (GSV) model of dry soil to represent the reflectance properties of dry soil. Then, we coupled these dry soil vectors with the MARMIT-2 model to propose the GSV + MARMIT-2 model. Finally, we assessed the accuracy of all three models using a wet soil database. The main conclusions of this study include: 1) the dry soil spectral vectors from the GSV model demonstrated high accuracy in describing the reflectance properties of dry soil, achieving an$R^{2}$of 0.988 and a root mean square error (RMSE) of 0.016. 2) All three soil models exhibited high fitting accuracy for the wet soil database ($R^{{2}} = \sim 0.992$and RMSE$= \sim 0.012$). Compared to the GSV and MARMIT-2 models, the GSV + MARMIT-2 model showed slightly improved accuracy under different soil moisture content (SMC) conditions. This study developed a more versatile and flexible soil model framework as it directly integrates the dry soil spectral vectors from the GSV model into the MARMIT-2 model. This coupling significantly expanded the applicability and improved the stability of the MARMIT-2 model.

In this work, we address a planar non-prehensile sorting task. Here, a robot needs to push many densely packed objects belonging to different classes into a configuration where these classes are clearly separated from each other. To achieve this, we propose to employ Monte Carlo tree search equipped with a task-specific heuristic function. We evaluate the algorithm on various simulated and real-world sorting tasks. We observe that the algorithm is capable of reliably sorting large numbers of convex and non-convex objects, as well as convex objects in the presence of immovable obstacles.

Moving a human body or a large and bulky object may require the strength of whole arm manipulation (WAM). This type of manipulation places the load on the robot's arms and relies on global properties of the interaction to succeed- rather than local contacts such as grasping or non-prehensile pushing. In this paper, we learn to generate motions that enable WAM for holding and transporting of humans in certain rescue or patient care scenarios. We model the task as a reinforcement learning problem in order to provide a robot behavior that can directly respond to external perturbation and human motion. For this, we represent global properties of the robot-human interaction with topology-based coordinates that are computed from arm and torso positions. These coordinates also allow transferring the learned policy to other body shapes and sizes. For training and evaluation, we simulate a dynamic sea rescue scenario and show in quantitative experiments that the policy can solve unseen scenarios with differently-shaped humans, floating humans, or with perception noise. Our qualitative experiments show the subsequent transporting after holding is achieved and we demonstrate that the policy can be directly transferred to a real world setting.