A circular honeycomb lower limb protection device was proposed to reduce the damage to the occupant’s lower limbs under a vehicle under-belly blast. First, a local equivalent model of the occupant-restraint system was established, and drop impact tests and theory were used to validate the model’s accuracy. Then, under the same mass, the protection performance of eight lower limb protection devices using circular honeycomb, hexagonal honeycomb, reentrant honeycomb, etc., as sandwiches were compared. It was found that the lower limb protection device with a circular honeycomb sandwich provides the best defence for lower limbs. Subsequently, the effect of gradient structure settings and dimensional parameters on the protection performance of circular honeycomb lower limb protection devices was investigated. Finally, multi-objective optimization has been carried out to further improve its protection performance. The results indicated that the protection performance of the lower limb protection device could be effectively improved by reasonably selecting the cell arrangement, gradient type and gradient interval of the honeycomb sandwich. When the optimized lower limb protection device was employed, the occupant's left and right lower tibial peak forces were decreased to 3.67 kN and 3.55 kN, respectively. Compared with the initial design, the optimized lower limb protection device reduced the left and right lower tibial peak forces by 21.1% and 23.8%, respectively, and the total mass decreased by 51.5%.