Patients with stroke can experience significant gait impairments, limiting their mobility and quality of life. Robotic exoskeletons offer a promising avenue for gait rehabilitation, but simple-to-use and effective control strategies are crucial for clinical settings. This paper presents a pilot study investigating the feasibility and potential benefits of a data-driven impedance control approach for lower-limb exoskeletons in two participants with stroke. Participants underwent a training session with the exoskeleton during which kinematic and EMG data were collected. Preliminary results from these experiments suggest that the data-driven impedance controller was able to provide appropriate assistance during five distinct ambulation modes (level-ground walking, ramp ascent/descent, stair ascent/descent), increasing stride length, foot clearance, reducing gait speed variability, while maintaining the user active in the exoskeleton (similar or higher EMG activation), in comparison to natural walking.