For space-ground heterogeneous multi-agent systems with state constraints, an adaptive finite time-constrained formation control method is proposed Considering the consistency requirement of multi-agent systems and the different state constraints of unmanned aerial vehicles and unmanned vehicles, the state constraint problem is analyzed as a state constraint problem, and the state constraint is transformed into an unconstrained state by a nonlinear transformation function. Based on the estimation of the uncertain nonlinear terms in the system, an adaptive neural network controller for each agent is designed and constructed by the backstepping method. Combined with the corresponding control law, all heterogeneous agents can achieve finite-time convergence to the desired formation trajectory. Finally, the stability and convergence of the multi-agent system are analyzed.