We describe a distributed and autonomous technique for dynamic gait adaptation for a chain-type, modular self-reconfigurable robot (MSR) using a fuzzy logic based, closed-loop controller. To maneuver itself, each module of the MSR is provided with a set of basic or fundamental gaits within a gait control table(GCT). A relevant problem in locomotion of a chain-type MSR is how to coordinate the gait of the individual modules with each other so that the desired locomotion of the MSR can be achieved. To address this problem, our proposed controller maps the inputs from the sensors of each module to an appropriate gait for the module determined from the goal and position of the module in the configuration, using a fuzzy technique. An inertial measurement unit (IMU) is used to close the loop between the goal and the module. We have verified the operation of our controller on a simulated 3-D model of an MSR called ModRED within the Webots robot simulator and also implemented it on the physical ModRED MSR. Our results illustrate that our controller can successfully adapt ModRED's locomotion by dynamically combining basic gaits from the individual modules in the configuration, regardless of the number of modules in the configuration and in the presence of noisy sensor inputs.