This article proposes a novel prescribed performance-based neural adaptive control scheme for robot manipulators including motor dynamics under model uncertainties without velocity, acceleration, and input current measurements. The prescribed performance function approach is used to transform a constrained tracking problem of the robot model including motor dynamics into an unconstrained third-order error model in Euler-Lagrange form which inherits all properties of the robot dynamics. Then, a projection-type neural adaptive PID <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> controller (a PID controller with the second-order derivative) in conjunction with a velocity-acceleration observer is proposed. Lyapunov's direct method is used to prove that the tracking and state observation errors are semiglobally uniformly ultimately bounded and converge to a small ball around the origin with a prescribed overshoot/undershoot, convergence rate, and final tracking accuracy. Finally, simulation, experimental results on a SCARA robot and comparative studies verify that the proposed controller is effective for the joint position trajectory tracking of robot manipulators in the industrial automation with minimum measurement and hardware requirements.