The global environmental crisis demands innovative technological solutions that can help humanity monitor, protect, and restore our natural world. Outdoor robotics and artificial intelligence offer promising approaches to address these challenges by enabling autonomous systems to operate effectively in complex natural environments with minimal human intervention. We examine the state of sensing technologies and perception algorithms for unstructured natural environments, highlighting recent advances in scene analysis, semantic segmentation, simultaneous localisation and mapping (SLAM), and traversability analysis. The chapter also explores perception in the context of multi-robot systems approaches to enhance environmental monitoring and management capabilities. Beyond technological aspects, we discuss challenges and opportunities that span computational resource management, end-user adoption, and ethical considerations. By assessing the current scientific and technological landscape, we identify gaps between research and practical implementation across environmental domains. The chapter concludes with a roadmap for sustainable environmental robotics that aligns with global climate action targets and sustainable development goals, featuring case studies of the Forestry Robotics at the University of Coimbra (FRUC) research group and the Robotics and AI for a Sustainable Environment (RAISE) initiative at Nottingham Trent University. Throughout, we position robotics and AI as critical tools in addressing pressing environmental challenges whilst emphasising the importance of responsible development and deployment.