Apple researcher on reset-free RL, reversibility benchmarks, and skill-library learning for robot manipulation
2
papers here
1
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
1
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-17) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.
Research profile
Output—
Not yet scored: the OpenAlex snapshot shows only an h-index of 1 and an i10-index of 0, too thin to characterize output volume reliably. This axis would measure total indexed publications, venue distribution, and yearly productivity across sources.
Impact—
Not yet scored: citation data is insufficient in the snapshot (h-index 1, i10-index of 0), so scholarly impact cannot be estimated. This axis would measure citation counts, citing works, and influence on reset-free RL and robot-learning research.
Focus100/100
Score 100/100. Both reviewed papers sit in reinforcement learning / robotics, the only field in the distribution, and the topic list is tightly concentrated: 'Reinforcement Learning in Robotics' and 'Robot Manipulation and Learning' are the only topics appearing twice, while recommender systems, AI-based planning, and neural-network applications appear once each. The agenda is coherently centered on reset-free RL and skill libraries for manipulation.
Collaboration14/100
Score 14/100, reflecting a minimal recorded network: the snapshot lists exactly 1 distinct co-author, with Riyaaz Shaik as the recurring partner on both 2026 arXiv papers. On each, Venkataraman sits in the last-author position without being the corresponding author. The pattern suggests a compact two-person collaboration, typical of an industry research setting, rather than the multi-institution networks this axis usually rewards.
Activity40/100
Score 40/100. Counts-by-year data shows a sparse trail — one indexed work in 2022 and one in 2025 — followed by two works in 2026, giving 3 works in the last three years. Both cataloged 2026 papers are arXiv preprints, indicating a recent, preprint-first burst of activity after a quiet early record rather than sustained high-volume publishing.
Reproducibility55/100
Score 55/100, averaged over only 1 scored paper, so treat it as provisional. Both 2026 papers carry 'medium' readiness: REVERSAL-BENCH is judged well-designed and causally controlled but simulation-only with thin quantitative reporting, while REFACTOR-VLA earns praise for honest reporting yet depends on task-label supervision. Benchmark and library artifacts bode well, but full replication paths are not yet demonstrated.
Research summary
Chandru Venkataraman is a researcher at Apple (United States) working at the intersection of reinforcement learning and robot learning. This profile is name-resolved from OpenAlex bibliographic data and remains early-stage: it records an h-index of 1, an i10-index of 0, a sparse counts-by-year trail (one work in 2022, one in 2025, two in 2026), and 3 works in the last three years. The small footprint belies a coherent agenda visible in two 2026 arXiv papers in the Evolve catalog. REVERSAL-BENCH introduces a reversibility axis (rho) and a simulator-based reset oracle across five physics engines, documenting a sharp 'cliff' for reset-free RL — a benchmark contribution to safe, autonomous reinforcement learning. REFACTOR-VLA applies wake/sleep learning to build typed motor-program skill libraries for vision-language-action models, beating baselines 4/4 on LIBERO NMI, though Evolve notes the 'unsupervised' framing is strained by reliance on task-label supervision. Venkataraman is last author on both (non-corresponding), working with Riyaaz Shaik. Topics concentrate tightly on RL in robotics and robot manipulation; both papers rate medium reproducibility readiness (mean 55/100 from a single scored paper). Citation-based output and impact dimensions cannot yet be computed.
Collaboration pattern
The recorded collaboration footprint is narrow: OpenAlex lists exactly 1 distinct co-author, and Riyaaz Shaik is the recurring partner on both 2026 arXiv papers in the catalog. Venkataraman takes the last-author position on each without being the corresponding author — a compact two-person setup consistent with an industry research context at Apple. The data shows no evidence yet of multi-lab or cross-institution collaboration.
Research keywords
Highlights
REVERSAL-BENCH: mapping the reset-free RL cliff
REFACTOR-VLA: wake/sleep skill libraries for VLAs
Timeline
- 2022Earliest trace in the snapshot: OpenAlex counts-by-year records one indexed work in 2022.
- 2025One further indexed work appears in 2025, keeping the recorded early output sparse.
- 2026REVERSAL-BENCH (arXiv) introduces a reversibility axis and a reset oracle across five physics engines; Venkataraman is last author.
- 2026REFACTOR-VLA (arXiv) presents wake/sleep learning of typed motor-program skill libraries for vision-language-action models; again last author.
Papers in the Evolve catalog
Frequent co-authors
About this profile
This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.
- All bibliometrics (h-index 1, i10-index 0, counts by year, topics, 1 distinct co-author) come from an OpenAlex snapshot; identity is name-resolved from public bibliographic data with no ORCID, and has not been independently verified.
- Output and impact dimensions are unscored because the snapshot lacks sufficient citation data.
- The reproducibility score (55/100) averages a single scored paper; both cataloged works are arXiv preprints rated medium.
- Coverage is limited to the Evolve-reviewed papers and this snapshot; works not indexed by OpenAlex are not reflected.