This study presents a liquid crystal elastomer (LCE) soft robot capable of achieving self-sustained multimodal locomotion. The robot can switch between different motion modes, such as rolling and jumping, through substrate adhesion or remote light stimulation. The LCE is mechanically trained to undergo repeated snapping actions to ensure self-sustained rolling motion in a constant thermal gradient. By adjusting the substrate adhesion, the robot can reversibly transition between rolling and jumping modes. Additionally, the rolling motion can be manipulated in real time through light stimulation to perform various complex motions, including turning, decelerating, stopping, backing up, and steering around obstacles. The principle of introducing an on-demand gate control offers a new approach for designing future autonomous soft robots. The LCE robot is synthesized using four monomer precursors and undergoes mechanical training through a two-stage process involving stretching and UV curing. The mechanical training enables the LCE to exhibit self-sustained rolling motion under a thermal field. The robot can also perform autonomous jumping motions by manipulating energy accumulation during the rolling process. The LCE robot demonstrates continuous rolling motion under a constant thermal field, with the motion direction being random. The speed of the robot is influenced by factors such as sample width, length, and hotplate surface temperature. The robot can also perform complex motions in response to light stimulation, including reversing direction, stopping, decelerating, turning, and changing from straight-line to circular motion. The LCE robot is capable of performing a variety of complex motions, including navigating around obstacles and exhibiting thermotaxis. The robot can also be released from a cold trap using light stimulation. The study highlights the potential of combining self-sustained motion with on-demand control methods for designing future soft robots. The LCE robot demonstrates the ability to perform complex functions that are challenging for known self-sustained systems. The robot can operate in both air and water environments, and its performance is not significantly affected by the surrounding medium. The study also discusses the limitations of the LCE robot, including its requirement for high temperatures and the need for further optimization to improve its performance in various environments. The results demonstrate the potential of LCE robots for applications in autonomous robotics and soft robotics.This study presents a liquid crystal elastomer (LCE) soft robot capable of achieving self-sustained multimodal locomotion. The robot can switch between different motion modes, such as rolling and jumping, through substrate adhesion or remote light stimulation. The LCE is mechanically trained to undergo repeated snapping actions to ensure self-sustained rolling motion in a constant thermal gradient. By adjusting the substrate adhesion, the robot can reversibly transition between rolling and jumping modes. Additionally, the rolling motion can be manipulated in real time through light stimulation to perform various complex motions, including turning, decelerating, stopping, backing up, and steering around obstacles. The principle of introducing an on-demand gate control offers a new approach for designing future autonomous soft robots. The LCE robot is synthesized using four monomer precursors and undergoes mechanical training through a two-stage process involving stretching and UV curing. The mechanical training enables the LCE to exhibit self-sustained rolling motion under a thermal field. The robot can also perform autonomous jumping motions by manipulating energy accumulation during the rolling process. The LCE robot demonstrates continuous rolling motion under a constant thermal field, with the motion direction being random. The speed of the robot is influenced by factors such as sample width, length, and hotplate surface temperature. The robot can also perform complex motions in response to light stimulation, including reversing direction, stopping, decelerating, turning, and changing from straight-line to circular motion. The LCE robot is capable of performing a variety of complex motions, including navigating around obstacles and exhibiting thermotaxis. The robot can also be released from a cold trap using light stimulation. The study highlights the potential of combining self-sustained motion with on-demand control methods for designing future soft robots. The LCE robot demonstrates the ability to perform complex functions that are challenging for known self-sustained systems. The robot can operate in both air and water environments, and its performance is not significantly affected by the surrounding medium. The study also discusses the limitations of the LCE robot, including its requirement for high temperatures and the need for further optimization to improve its performance in various environments. The results demonstrate the potential of LCE robots for applications in autonomous robotics and soft robotics.