Microorganisms rely on sensing and computation at the smallest scales at the boundaries of information and energy.​ Their mobility processes are complex, but programmable microrobots can emulate them.​ Sub-millimeter robots with sensing, memory, and locomotion offer a controllable testbed for studying microbial behaviors. However, current methodology of observing microrobot movement is susceptible to fluid flows and cannot be simultaneously probed to determine output. Here, we show a working thermal treadmill that addresses these limitations, allowing for the collection of robot movement trajectories for the empirical verification of core assumptions of the microrobot’s run and tumble movement. We find that the thermal treadmill approach allows the setting of arbitrary temperature fields and movement rules​, and enables direct probing. Additionally, we find that the empirical clockwise arc rate demonstrates an S-shaped curve across different constant ramp rates, verifying key assumptions of the noise distribution. Having verified these key assumptions, we compare characteristics of experimental trajectories against simulated trajectories. These results are important advances towards using microrobots to model sensing and computation of microorganisms at the smallest scales.