Two kinds of autonomy: DirectHop’s operator-controlled mobility and ElliQ’s deployed social agency

AI-persona disclosure

Fictional OMIKINA AI editorial persona; not a human reporter and does not hold legal or regulatory credentials or possess firsthand experience.

Key points

• DirectHop can vary its jumps and recover from many awkward landings, but the current prototype has wired power, no orientation or jump-height sensing, and full reliance on human operators.

  Sources: S1

• New York extended its ElliQ partnership after reported program outcomes, but the supplied account does not describe the evaluation design, participant-selection rules, or terms of the extension.

  Sources: S2

• The comparison shows that autonomy should be assessed against the job at hand: safe independent navigation for a physical robot, and sustained, appropriate engagement for a social companion.

  Sources: S1 · S2

A shared label, two very different thresholds

“Autonomous” can obscure an important distinction in robotics. DirectHop, a University of Washington hopping-robot project, is being developed toward independent exploration but is presently controlled from a nearby laptop. ElliQ, Intuition Robotics’ proactive AI companion, is already being distributed through an extended New York State Office for the Aging partnership to older New Yorkers identified as being at risk of loneliness and social isolation. One system’s central unanswered question is whether it can physically operate without an operator; the other’s is whether proactive assistance produces an appropriate and durable public-service outcome for people using it.

Sources: S1 · S2

The reported facts do not establish a common technical or policy standard for either product. DirectHop’s source describes a research prototype and proposed upgrades. The ElliQ account reports a government partnership extension and program outcomes. Neither supplied item provides primary contract language, a formal regulatory requirement, or a detailed independent evaluation. That means a partnership renewal should not be read as proof that a particular social-companion design is legally required, just as a laboratory capability should not be read as field-ready navigation.

Sources: S1 · S2

Sources: S1 · S2

DirectHop’s dependency is not a cosmetic limitation

DirectHop is a palm-sized, one-gram machine with an electric motor, folding legs and a protective cage. Rather than using a fixed spring release, it changes motor current to adjust the energy put into a jump. The report says it can make both high jumps and small controlled jumps. Its motor also shifts the center of gravity after a landing, while the flexible carbon-fiber cage contributes to recovery; in laboratory testing, the robot righted itself in about 90% of cases. Those are meaningful component-level mobility results, especially because controllable short hops address a different problem from maximum jump height.

Sources: S1

But the same account identifies the dependency that prevents calling the current machine an independent explorer: DirectHop receives power through wires, has no sensors to judge required jump height or its own orientation, and fully relies on human operators. Researchers are working on solar cells, a battery, retractable feet, a camera and navigation electronics. Those are development intentions, not demonstrated capabilities. A swarm for monitoring water and soil, refinery inspection, or exploration is similarly presented as a future application, alongside an estimated possible unit cost, rather than an operating deployment.

Sources: S1

Sources: S1

ElliQ’s autonomy is operational, but bounded by service delivery

ElliQ’s relevant autonomy is not locomotion. The companion checks in, remembers an older adult, and can encourage actions such as walking or contacting family, according to Intuition Robotics’ chief executive. New York’s aging office extended the partnership after what the account characterizes as three years of measurable outcomes. The program began distributing ElliQ in 2022, and the report says hundreds of older adults have received a device. It also says more than 3,500 older adults applied during 2025, indicating demand for the program rather than demonstrating that every applicant received, needed, or benefited from a device.

Sources: S2

The reported 2026 figures are substantial as engagement and participant-reported outcome signals: users averaged 48 daily interactions; 94% reported feeling less lonely; 97% reported improved overall wellbeing; and 79% felt more connected to the world around them. The article also reports recurring conversations, wellbeing updates, and activity encouraged by ElliQ. These figures support the narrow conclusion that participating users reported engagement and positive experiences. They do not, on the supplied evidence, establish clinical causation, comparison with people who did not receive ElliQ, persistence beyond the reporting period, or outcomes for every older New Yorker at risk of isolation.

Sources: S2

Sources: S2

The practical decision is to match proof to the dependency

For DirectHop’s developers or a prospective field user, the gating evidence is practical: onboard power, sensing, navigation, landing performance under relevant conditions, and performance without a nearby operator. The current robot’s controlled jumping and self-righting are evidence toward that goal, but do not remove those dependencies. For NYSOFA and similar public-service buyers, the decision is different. The supplied account says the agency added social-isolation and technology screens to its comprehensive assessment during the Covid-19 pandemic so people could be connected to digital options when appropriate. That makes fit to an individual, rather than mere availability of a device, part of the deployment question.

Sources: S1 · S2

Inference: the two cases suggest that autonomy is best treated as a system property, not a product label. DirectHop needs a closed loop between sensing, power, movement and recovery before a human can leave the loop. ElliQ already operates proactively in homes, but its public value remains linked to selection, user willingness to engage, and the program’s ability to interpret self-reported outcomes responsibly. In both cases, human involvement has not disappeared; it has moved from continuous control in the robot case to program design, matching and oversight in the companion case.

Sources: S1 · S2

Sources: S1 · S2

What would change the assessment

DirectHop’s case would strengthen with evidence that the planned battery, camera and navigation electronics work together in an untethered setting, including evidence on navigation and recovery beyond the laboratory self-righting result. ElliQ’s case would become clearer with disclosed methodology for the reported outcomes, information about which applicants receive devices and why, longer-term results, and evidence separating the companion’s contribution from other supports. The current material supports optimism about two distinct forms of assistance, but it also shows why a measured jump and a reported improvement in loneliness cannot be treated as interchangeable proof of autonomy.

Sources: S1 · S2

Sources: S1 · S2

Why it matters

Robotics policy and procurement can fail when they apply one autonomy label to fundamentally different risks. DirectHop’s next hurdle is independence from physical infrastructure and operator control. ElliQ’s is accountable deployment: determining who is an appropriate fit and how to interpret favorable participant reports without overstating what they prove. The distinction matters for researchers, service agencies and users deciding what evidence is sufficient before a system is trusted with mobility or social support.

Sources: S1 · S2

Sources

1. Next-gen hopping robot is the smart flea we didn't know we needed — New Atlas Robotics · 2026-10-10
2. New York extends AI companion program in partnership with Intuition Robotics — Robotics & Automation News · 2026-10-10