Inova and Lattice demonstrate low-latency motor control for humanoid robots
Inova Semiconductors and Lattice Semiconductor have demonstrated a distributed motor control platform for humanoid robots, using FPGA technology to synchronise the movement and sensing of an 18-motor robotic hand.
Showcased at electronica 2026, the development platform combines Inova’s APXpress high-speed networking technology with low-power Lattice CertusPro-NX and CrossLink-NX FPGAs. The German Aerospace Centre (DLR), through its Institute of Robotics and Mechatronics, is the scientific partner in the project.
The system connects all 18 motors, joint sensors and a tactile array comprising 10 sensors through a single deterministic network. This architecture enables the robotic hand to sample sensor data and adjust motor currents within the same 10kHz control cycle.
Using APXpress and field-oriented control (FOC) algorithms developed by DLR, the demonstrator achieves communication latency of less than 10µs between sensor sampling and the transmission of new target values. All motor axes operate synchronously, reducing control delays that can affect movement accuracy and stability.
Unlike conventional cyclic control systems, which may take one or two additional cycles to apply motor adjustments, the same-cycle approach helps improve responsiveness. This is particularly important for stable impedance control and safe human-robot interaction.
The motor controllers use Lattice FPGAs to implement FOC locally at each motor, with control running at a higher frequency than the network update rate.
The APXpress network also carries joint-angle and tactile sensor data. Its bandwidth and scheduling capabilities allow higher-volume tactile information to be transferred without interfering with time-critical motor control messages.
A daisy-chain network topology reduces cabling compared with conventional star architectures, helping minimise wiring weight and free up space around the hand’s mechanical joints. The architecture can also accommodate additional sensor inputs, including cameras and radar.
Inova is developing the silicon for its modular APXpress network architecture, with each node based on Lattice CertusPro-NX or CrossLink-NX FPGAs. Lattice is providing application engineering support, while DLR contributes its robotics and manipulation expertise.
fpga-motor-control-humanoid-robot-hand-inova-lattice
