In October 2026, the TRANS Research Group developed a study on task-level energy assessment and predictive risk evaluation for embodied mobile robots under dynamic operating conditions. This work proposes a physics-constrained multi-task learning framework that integrates task-available energy (TAE), future energy consumption, minimum battery voltage, and maximum battery temperature prediction. Validated on 85 days of real-world operational data comprising 61,178 task fragments across five locomotion modes, the proposed model achieves NRMSE values of 1.34%, 1.60%, 0.68%, and 0.59% for the four prediction targets, respectively. Compared with the strongest energy-related baselines, it reduces NRMSE for TAE and future energy consumption by 72.4% and 65.6%, respectively. The predictive relative risk margin further achieves 94.01% risk-level agreement, supporting interpretable energy assessment and risk-aware operational decision-making for mobile robots.
Lv, Z., Ou, S. S., He, S., Jiang, Y., Jing, H., & Zhang, C. (2027). Physics-constrained multi-task energy consumption prediction for the safe operation of embodied mobile robots. Applied Energy, 427, 129151. https://doi.org/10.1016/j.apenergy.2026.129151