FPGAs offer hardware-level protection for robotic systems
Eric Sivertson of Lattice Semiconductor argues that humanoid robots must be both safe and secure. He explains that software-based systems are vulnerable to cyberattacks, and FPGAs provide a hardware layer to lock down the system. The discussion highlights the need for security in physical AI applications.
Sivertson, speaking on a podcast, contrasts traditional stationary robots with mobile humanoids, noting that safety protocols rely on data that could be tampered with. He highlights that a mass compromise of humanoid robots could turn them into physical threats due to their strength.
To counter these risks, he emphasizes mutual attestation between components and clouds, along with digital signatures to verify authenticated code. This hardware-based approach aims to merge safety and security, which have historically been treated as separate domains.
The integration of hardware-level security in robotic systems could significantly alter the risk landscape for industries deploying physical AI. Manufacturers and operators of humanoid robots may face lower risks of catastrophic cyber-physical attacks, while cybersecurity professionals might need to adapt to hardware-centric defenses. Society at large could benefit from safer autonomous machines, though the added complexity may raise costs or slow adoption. Ultimately, this approach could set a new baseline for trust in robotic infrastructure.