Benchmark discovery
Map appropriate protocols to your robot class, task, and evaluation constraints.
Discuss a discovery scope →Independent · Source-backed · Protocol-aware
The independent index of robotics benchmarks
Compare simulation, real-robot, and compute benchmarks in one place. Every entry includes source links, a verification date, and confidence notes.
10 reviewed starters40 primary sources0 synthetic rankings
SimulationReal robot
TaskSystem
Manipulation
Navigation
Locomotion
Compute
EvidenceProtocolMetric
Evaluation services · Scope first
Make your next robotics decision evidence-led
Benchmark discovery, capability assessments, and technical intelligence for robotics teams and buyers. Every engagement is scoped to the system, environment, and evidence available.
Capability assessment
Define measurable evaluation criteria, traceable sources, and explicit limitations.
Request an assessment →Technical intelligence
Translate vendor claims and research evidence into a decision-ready briefing.
Request a briefing →A shared comparison language
Built for decisions, not hype
Consistent schema
Every benchmark is described with the same fields, so project-specific language does not hide meaningful differences.
Transparent provenance
Material claims link to official project pages or papers and carry an explicit verification state and timestamp.
Practical guidance
Category hubs and selection guidance help teams match evaluation resources to their actual task, system, and setting.
Primary-source reviewed
Recently verified entries
| Name | Domain | Setting | Metric state | Verified |
|---|---|---|---|---|
| Open X-Embodimentdataset resource | manipulation | real robot | No universal metric | Aug 20, 2026 |
| LIBERObenchmark suite | manipulation | simulation | Forward Transfer | Aug 20, 2026 |
| RLBenchbenchmark suite | manipulation | simulation | Task success rate | Aug 20, 2026 |
| RobotPerfbenchmark suite | systems_compute | compute only | No universal metric | Aug 20, 2026 |
| RoboChallenge Table30challenge | manipulation | real robot | End-to-end task success rate | Aug 20, 2026 |
| Habitatplatform framework | multi_domain | simulation | No universal metric | Aug 20, 2026 |
Browse by evaluation domain
Start with the problem you need to measure
07
Manipulation
Task, skill, and generalization evaluations for robot interaction with objects and environments.
Explore category 01Navigation
Embodied navigation, mapping, exploration, and collision-aware planning resources.
Explore category 00Locomotion
Control and mobility evaluations across terrain, morphology, and operating conditions.
Explore category 00Compute
Performance, latency, energy, and system-level benchmarking for robotics workloads.
Explore category 00Multi-domain
Resources spanning multiple robotics domains, embodiments, or evaluation settings.
Explore category 00Challenges
Competitions and real-world evaluation programs with protocol- and stage-specific results.
Explore categoryMethodology
Sources are official project pages, repositories, rules, and primary papers. Demo rows are labeled. No benchmark is ranked or endorsed.
Evidence cutoff
August 20, 2026
A better evaluation starts with scope