This paper presents ReachBot, a robot designed to navigate and manipulate in challenging environments such as lunar and Martian caves and lava tubes. Traditional robots are not suitable for these environments due to their inaccessible terrain. ReachBot uses extendable booms as appendages to manipulate itself with respect to irregular rock surfaces. The booms terminate in grippers equipped with microspines, allowing ReachBot to achieve force closure in enclosed spaces. To propel ReachBot, a contact-before-motion planner is used, which utilizes internal force control to keep the booms in tension. Motion planning also depends on finding and executing secure grips on rock features. A Monte Carlo simulation is used to inform gripper design and predict grasp strength and variability. Additionally, a two-step perception system is used to identify possible grasp locations. The robot was tested in a lava tube in the Mojave Desert, confirming its ability to find many targets for secure grasps. The paper also discusses the design of the grippers, the perception system, and the motion planning strategy. The results show that ReachBot can navigate and manipulate in difficult terrain, demonstrating the potential of a full-scale ReachBot system. The paper also discusses the challenges and areas of future work in motion planning, mechanism design, and perception.This paper presents ReachBot, a robot designed to navigate and manipulate in challenging environments such as lunar and Martian caves and lava tubes. Traditional robots are not suitable for these environments due to their inaccessible terrain. ReachBot uses extendable booms as appendages to manipulate itself with respect to irregular rock surfaces. The booms terminate in grippers equipped with microspines, allowing ReachBot to achieve force closure in enclosed spaces. To propel ReachBot, a contact-before-motion planner is used, which utilizes internal force control to keep the booms in tension. Motion planning also depends on finding and executing secure grips on rock features. A Monte Carlo simulation is used to inform gripper design and predict grasp strength and variability. Additionally, a two-step perception system is used to identify possible grasp locations. The robot was tested in a lava tube in the Mojave Desert, confirming its ability to find many targets for secure grasps. The paper also discusses the design of the grippers, the perception system, and the motion planning strategy. The results show that ReachBot can navigate and manipulate in difficult terrain, demonstrating the potential of a full-scale ReachBot system. The paper also discusses the challenges and areas of future work in motion planning, mechanism design, and perception.