• Robotic algorithms are evolving from traditional to hierarchical and end-to-end models, posing challenges for balanced system-wide performance.
• Algorithm-hardware co-design is essential to analyze computational properties and drive both algorithm optimization and hardware innovation.
• Recent works demonstrate co-design across robotic and embodied AI algorithms and computing hardware, achieving high performance and energy efficiency.
• Future research must adapt computing platforms to rapid embodied AI evolution and leverage emerging hardware for end-to-end inference improvements.