The method builds information bidirectional fusion filters on the basis of different error divergence of the foot-mounted inertial navigation system and dead reckoning / simultaneous localization and mapping（ SLAM）navigation system. The zero-velocity update（ ZUPT） of biped robot is used to reduce the error of navigation system

while the SLAM algorithm of wheeled robot is used to correct the positioning error

which both improve the positioning and heading precision of two navigation systems synchronously. Experimental results show that

the positioning error of navigation system is about 2. 3 % of the walking distance

which can effectively improve the navigation performance of the robot group in the environment without GNSS.