Abstract
The growing demand for energy-efficient robotic systems has driven research on energy-conscious trajectory planning and control for manipulators, yet findings remain scattered across methods and evaluation schemes. This systematic literature review aims to clarify how energy-conscious concepts are formulated and implemented in trajectory research and to classify the links between analytic–numeric optimization and reinforcement-learning-based approaches. Following PRISMA guidelines, we queried Scopus (Springer and IJRCS) for journal articles published between 2021 and August 2025 and identified 124 primary studies. A structured extraction form and a taxonomy scheme mapped each paper to four research questions: (i) energy formulation, (ii) trajectory methods, (iii) system models and evaluation setups, and (iv) research gaps and future directions. Synthesis combined descriptive statistics with matrix-based and qualitative analysis. Results show that explicit energy-conscious formulations (energy models, torque or jerk penalties, power limits) appear only in a subset of works, while most studies still optimize indirect quantities such as time, smoothness, or tracking error. One-DoF configurations are frequently used as controlled testbeds for dynamics- and analytics-based energy studies. Kinematic or trajectory-based analytics and dynamics dominate the corpus (proportions 0.77 and 0.60), whereas hybrid, numeric and heuristic, and ML- or RL-based methods are less prevalent. Among the 22 studies that explicitly address energy-conscious aspects, these proportions increase to 0.96 and 0.86, indicating that analytic–dynamic formulations currently form the backbone of energy-efficient trajectory research while leaving substantial room for deeper integration between trajectory optimization and learning-based control. The review outlines priorities for multi-DoF energy models, real-time control, and energy-aware RL.