Tell Robot What Not to Do: A Negation Understanding Perspective
Overview
NegaAlign improves negated instruction success rates on simulated and real-world tasks, reaching 88.45% on NegaBench and 88.8% in real-world scenarios using a parameter-efficient framework for vision-language-action models.
Generated from attributed reports · Updated 1 hours ago
Event evidence and corrections
0 attributed source owners. Ownership does not establish independent confirmation. Quantities are reported separately and are never added together.
Reported quantity · Reported success rate: 88.45 percent · Basis not reported · Differing source assertions
“enchmark spanning 10 scenarios across five domains for systematically evaluating manipulation under negated constraints. Experiments across GR00T, $\pi_0$, and $\pi_{0.5}$ demonstrate consistent improvements in negated instruction following. With 11.6M trainable parameters, NegaAlign increases the negated-instruction success rate of $\pi_{0.5}$ from 2.60% to 88.45% on NegaBench and from 12.4% to 88.8% on real-world t”
Exact source · revision 1Source owner not reported
Reported quantity · Reported success rate: 88.8 percent · Basis not reported · Differing source assertions
“enchmark spanning 10 scenarios across five domains for systematically evaluating manipulation under negated constraints. Experiments across GR00T, $\pi_0$, and $\pi_{0.5}$ demonstrate consistent improvements in negated instruction following. With 11.6M trainable parameters, NegaAlign increases the negated-instruction success rate of $\pi_{0.5}$ from 2.60% to 88.45% on NegaBench and from 12.4% to 88.8% on real-world t”
Exact source · revision 1Source owner not reported
Reported quantity · Reported success rate: 2.6 percent · Basis not reported · Differing source assertions
“enchmark spanning 10 scenarios across five domains for systematically evaluating manipulation under negated constraints. Experiments across GR00T, $\pi_0$, and $\pi_{0.5}$ demonstrate consistent improvements in negated instruction following. With 11.6M trainable parameters, NegaAlign increases the negated-instruction success rate of $\pi_{0.5}$ from 2.60% to 88.45% on NegaBench and from 12.4% to 88.8% on real-world t”
Exact source · revision 1Source owner not reported
Reported quantity · Reported success rate: 12.4 percent · Basis not reported · Differing source assertions
“enchmark spanning 10 scenarios across five domains for systematically evaluating manipulation under negated constraints. Experiments across GR00T, $\pi_0$, and $\pi_{0.5}$ demonstrate consistent improvements in negated instruction following. With 11.6M trainable parameters, NegaAlign increases the negated-instruction success rate of $\pi_{0.5}$ from 2.60% to 88.45% on NegaBench and from 12.4% to 88.8% on real-world t”
Exact source · revision 1Source owner not reported
Reported quantity · trainable parameters: 11600000 other · Basis not reported
“enchmark spanning 10 scenarios across five domains for systematically evaluating manipulation under negated constraints. Experiments across GR00T, $\pi_0$, and $\pi_{0.5}$ demonstrate consistent improvements in negated instruction following. With 11.6M trainable parameters, NegaAlign increases the negated-instruction success rate of $\pi_{0.5}$ from 2.60% to 88.45% on NegaBench and from 12.4% to 88.8% on real-world t”
Exact source · revision 1Source owner not reported
Reported quantity · number of scenarios: 10 other · Basis not reported
“enchmark spanning 10 scenarios across five domains for systematically evaluating manipulation under negated constraints. Experiments across GR00T, $\pi_0$, and $\pi_{0.5}$ demonstrate consistent improvements in negated instruction following. With 11.6M trainable parameters, NegaAlign increases the negated-instruction success rate of $\pi_{0.5}$ from 2.60% to 88.45% on NegaBench and from 12.4% to 88.8% on real-world t”
Exact source · revision 1Source owner not reported
Reported quantity · number of domains: 5 other · Basis not reported
“enchmark spanning 10 scenarios across five domains for systematically evaluating manipulation under negated constraints. Experiments across GR00T, $\pi_0$, and $\pi_{0.5}$ demonstrate consistent improvements in negated instruction following. With 11.6M trainable parameters, NegaAlign increases the negated-instruction success rate of $\pi_{0.5}$ from 2.60% to 88.45% on NegaBench and from 12.4% to 88.8% on real-world t”
Exact source · revision 1Source owner not reported
Report timeline
Follow attributed reports and material updates.
- arXiv Robotics — research abstractsTell Robot What Not to Do: A Negation Understanding Perspective
This research presents NegaAlign, a parameter-efficient framework that allows vision-language-action models to follow negated instructions by reshaping intermediate instruction representations and using teacher-guided alignment. The framework improves negated instruction success rates on both simulated and real-world tasks, achieving up to 88.45% success on NegaBench and 88.8% on real-world tasks.
Timezone · UTC
Article dates follow your selected timezone. Briefing editions use Hong Kong time (UTC+8).