In this paper, we propose a fully decentralized control strategy for collective payload transport in 3D space by a team of identical holonomic mobile robots in a microgravity environment. This task has potential space applications such as on-orbit assembly, debris removal, and planetary construction, which have to be performed in uncertain environments where inter-robot communication is unreliable or not available. Decentralized control schemes that rely on limited data and do not require inter-robot communication would enable teams of robots to achieve cooperative transport objectives in unknown environments. Existing decentralized strategies for collective payload transport require at least some of the following information: (1) the geometry and dynamics of the payload; (2) the payload's position and velocity over time; (3) the number and distribution of robots around the payload; (4) the vector from the payload's center of mass to each robot's attachment point; and (5) desired trajectories for the robots and payload. In this paper, we consider a team of robots that must stabilize a payload's center of mass to a target position without access to any of this information. The robots are assumed to be rigidly attached to the payload, e.g. via pincers, grippers, or magnets, and the only information provided to them is the target position in a global coordinate system that is shared by all the robots. The robots do not exchange information and only require measurements of their own positions and velocities. Moreover, the payload may have an arbitrary shape, without particular constraints on its geometry (e.g., convexity). Given these assumptions and objectives, and building on our prior work on collective transport strategies, we design a fully decentralized proportional-derivative controller that can be employed by the robots to drive the payload's center of mass to the destination. Using a Lyapunov stability argument, we analytically prove that under this controller, the payload's center of mass asymptotically converges to a small neighborhood of the target position. The radius of this neighborhood is the magnitude of the error between the target position and actual steady-state position of the payload's center of mass. We also investigate the effect of the distribution of robots around the payload on the magnitude of this error. We validate our theoretical results in MATLAB simulations of five collective transport scenarios with different payload shapes and variations in the distribution of robots around the payload.