- Ezirim Kelechi ThankGod*, Sani Abubakar Muhammed, Aniugo Victor Onyekachi, Nwaokolo Ikechukwu Frank, Obi Obichukwu Immanuel, Okoronkwo Iheanyi Chinedu & Okwuonu Stanley Chinedu
- *Mechatronics Engineering Department; Federal University of Technology, Owerri, Imo State, Nigeria
- DOI: 10.5281/zenodo.21906170
Unraveling the solutions to the numerous challenges
poised on mobile robot navigation in cluttered, dynamic environments remain an
ambiguous concern in robotics. This is in spite of resounding advances so far
recorded in key areas like sensing, perception, and control, as most commercial
platforms obviously still rely heavily on human teleoperation when operating
outside structured warehouses. Hence the, review synthesizes 2015–2026 research
on the transition from teleoperation to full autonomy for mobile robots in
cluttered indoor and outdoor settings. Using a PRISMA-guided literature search
across IEEE Xplore, ACM, and Scopus, we analyze 127 peer-reviewed studies
through the lens of Technology Readiness Levels [TRL] and the “Sense-Plan-Act”
autonomy stack. Findings reveal three dominant pathways among others to wit: 1.
learning-based end-to-end navigation, 2. modular classical pipelines with
learned components, and 3. hybrid human-in-the-loop autonomy. However, key
bottlenecks still persist in sim-to-real transfer, long-horizon planning under
occlusion, and certifiable safety. We therefore propose a four-stage pathway:
Teleoperated Baseline; Shared Autonomy; Conditional Autonomy and Full Autonomy
while specifically recommending it for datasets, benchmarks, and regulatory
testbeds. The review concludes that modular-hybrid architectures currently
offer the most feasible and flexible route to deployment, while end-to-end
learning remains vital for long-term robustness.

