Search for: All records

Robot skill learning and execution in uncertain and dynamic environments is a challenging task. This paper proposes an adaptive framework that combines Learning from Demonstration (LfD), environment state prediction, and high-level decision making. Proactive adaptation prevents the need for reactive adaptation, which lags behind changes in the environment rather than anticipating them. We propose a novel LfD representation, Elastic-Laplacian Trajectory Editing (ELTE), which continuously adapts the trajectory shape to predictions of future states. Then, a high-level reactive system using an Unscented Kalman Filter (UKF) and Hidden Markov Model (HMM) prevents unsafe execution in the current state of the dynamic environment based on a discrete set of decisions. We first validate our LfD representation in simulation, then experimentally assess the entire framework using a legged mobile manipulator in 36 real-world scenarios. We show the effectiveness of the proposed framework under different dynamic changes in the environment. Our results show that the proposed framework produces robust and stable adaptive behaviors. 

Abstract Arctic amplification is leading to increased terrestrial organic carbon (terrOC) mobilization with downstream impacts on riverine and marine biogeochemistry. To improve quantification and characterization of terrOC discharged to the Arctic Ocean, Yukon River delta samples were collected during three stages of the annual hydrograph (ascending limb/peak freshet, descending limb, late summer) and across a land‐to‐ocean salinity gradient (0.08–29.06 ppt). All samples were analyzed for dissolved organic carbon (DOC) concentration and lignin phenols to determine seasonal variability in riverine terrOC and salinity‐induced transformation of highly aromatic terrestrial compounds. Additionally, the relationship between lignin and absorbance at 350 and 412 nm was assessed to determine the feasibility of using optical proxies for accurate quantification, both seasonally and across expansive salinity gradients. Lignin phenols were highest during the ascending limb/peak freshet (0.58–0.97 mg/100 mg OC) when riverine DOC was dominated by young vascular plant sources, whereas lignin phenols were lower (0.15–0.89 mg/100 mg OC) and riverine DOC more variable in terrestrial source and diagenetic state during the descending limb and late summer. Across the sampled salinity gradient, there was disproportionate depletion of lignin (up to 73%) compared to DOC (up to 22%). Finally, while optical proxies can be used to quantify lignin within seasonal or spatial contexts, increased uncertainty is likely when expanding linear correlations across Arctic land‐ocean continuums. Overall, results indicate seasonal, spatial, interannual, and climatic controls that are amplified during high‐flow conditions and important to constrain when investigating Arctic terrOC cycling and land‐ocean DOC flux.