Jalapeños Linked to Minnesota Salmonella Outbreak: At least 110 Salmonella Javiana infections in Minnesota have been linked to jalapeños served at Chipotle restaurants, according to state health officials and the company. Of 84 patients interviewed, 75 reported eating at Chipotle before becoming ill, with symptom onset dates ranging from June 14 to July 14. Chipotle said its ingredient-tracing system identified jalapeños from a common lot as a potential source. The chain removed the peppers from affected restaurants and replaced them with jalapeños supplied by different growers. Minnesota epidemiologists cautioned that the contaminated produce may also have been served at other Mexican-style quick-service restaurants. Officials have not yet determined whether the outbreak is still ongoing, though they said Chipotle has cooperated with the investigation. The inquiry continues as investigators trace the jalapeños’ distribution. (CIDRAP)

Cyclospora Outbreak Spreads Across 15 States: A large US Cyclospora outbreak linked to iceberg lettuce from Taylor Farms de Mexico has expanded into six additional states, bringing the total affected to 15. Newly identified cases have been reported in Arkansas, Iowa, Missouri, Nebraska, New Hampshire and North Carolina. As of August 5, federal officials had received 10,468 confirmed reports of domestically acquired cyclosporiasis, with more than 12,500 additional illnesses awaiting confirmation or investigation. Michigan, the hardest-hit state, has reported two deaths involving patients with serious underlying health conditions, though deaths from cyclosporiasis are uncommon. The parasitic illness typically causes prolonged watery diarrhea, nausea, bloating and abdominal cramps. Officials cautioned that case totals may continue rising because symptom onset, testing and reporting can create delays of approximately six weeks. Taylor Farms recalled the implicated lettuce on July 17. (CIDRAP)

Senate Confirms Erica Schwartz to Lead CDC: The US Senate has confirmed Erica Schwartz, MD, MPH, JD, as director of the Centers for Disease Control and Prevention in a largely party-line 51–44 vote. Schwartz previously served 24 years in the US Public Health Service Commissioned Corps and was deputy surgeon general during President Donald Trump’s first administration. She inherits an agency that has lacked a confirmed leader since Susan Monarez was removed in August 2025. Schwartz faced criticism during her confirmation hearing for declining to say whether she would oppose administration efforts to restrict vaccines. She also appeared unfamiliar with several recent CDC policy changes. Support from Republican Senator Bill Cassidy helped secure her confirmation. Schwartz now confronts outbreaks, widespread measles transmission, major staffing reductions and calls to restore scientific credibility and stability at the agency. (cidrap.umn.edu)

Robots Learn to Protect Human Partners’ Backs: Researchers have developed a human-aware control system that allows a humanoid robot to adjust its movements while sharing a heavy load with a person. Tested on the ergoCub robot, the framework estimates its partner’s posture and biomechanical strain, then changes its own position to make the joint lifting task safer and more comfortable. The system combines whole-body robot control with models predicting stress on the human lower back. Experiments indicated that this adaptive assistance reduced potentially harmful loading compared with approaches that focused only on completing the task. The work could help collaborative robots move beyond rigidly programmed factory routines toward physically responsive teamwork in warehouses, construction sites, hospitals and eldercare. The broader challenge will be ensuring that biomechanical predictions remain reliable across people with different bodies, abilities and movement styles. (Tech Xplore)

Flapping Robots Recover From Sudden Disturbances: A new control method could make insect- and bird-inspired flying robots more stable when struck by gusts or other unexpected forces. Flapping-wing micro aerial vehicles offer maneuverability and potentially quieter flight than propeller-driven drones, but their rapidly oscillating wings create complicated dynamics that make disturbance recovery difficult. Researchers devised a strategy that accounts for these changing forces rather than treating the robot as a conventional aircraft. In tests, the controller helped a flapping-wing platform regain stability following abrupt movements resembling wind interference. More dependable control could eventually support tiny robots used for environmental monitoring, confined-space inspection, search operations and biological research. However, laboratory stabilization does not automatically translate to outdoor autonomy, where turbulence, obstacles, limited battery capacity and uncertain sensor readings may occur simultaneously. The study represents progress toward making biologically inspired flight practical outside controlled environments. (EurekAlert!)

Tiny Robot Carries Many Times Its Weight: Engineers have created an untethered, insect-scale robot that combines movement strategies inspired by fish and seals. The three-legged machine uses a piezoelectric driving leg that vibrates in two directions, allowing it to move forward and rotate with only one actuator. The complete wireless prototype measures 39 millimeters long, weighs 20.9 grams and includes its own battery and control electronics. During testing, it traveled at more than six body lengths per second while carrying 180 grams—over eight times its own weight. A larger battery supported approximately 70 minutes of continuous movement. The robot performed best on smooth, hard surfaces, so rough terrain remains a significant limitation. Nevertheless, its combination of speed, payload capacity, compactness and onboard power suggests potential uses in narrow-space inspection, equipment monitoring and small-object transport. Future versions may incorporate vision and closed-loop navigation. (Science Advances)

Expressive Robots Lose Trust More Dramatically: Socially expressive robots may initially engage people more effectively, but their mistakes also inflict greater damage on trust. In a Drexel University-led experiment, 50 adult men interacted with either an animated Pepper humanoid—making eye contact, gesturing and nodding—or a motionless version delivering identical speech. Researchers simultaneously measured brain activity, oxytocin, survey responses and whether participants followed the robot’s advice. When the expressive robot began interrupting and offering irrational suggestions, participants appeared to process its errors as social violations rather than simple mechanical glitches. Its influence over their decisions fell by more than half. Surprisingly, oxytocin increased as trust declined, suggesting that the hormone may sometimes signal heightened social vigilance rather than bonding. The results indicate that robot designers should prioritize competence and reliability before adding humanlike charm, especially in healthcare, education or customer-facing applications. (Drexel University)

Porous Feet Help Quadrupeds Walk More Efficiently: Researchers have designed lightweight, porous, 3D-printed feet that reduce the energy required by quadruped robots. Four-legged machines are increasingly considered for industrial inspection, disaster response and transportation over terrain that wheeled vehicles cannot easily cross. Their repeated lifting, swinging and placement of the legs, however, consume substantial power and limit operating time. The new foot structures appear to reduce unnecessary mass while preserving the mechanical behavior needed for stable ground contact. Lower leg inertia means motors expend less energy accelerating and decelerating each limb during a walking cycle. The approach is notable because it improves efficiency through physical design rather than relying entirely on larger batteries or more complicated control software. Future studies will need to assess durability, traction and performance on mud, rocks and loose soil. Even modest energy savings could meaningfully extend missions for untethered field robots. (Phys.org)

Muscle Radar Could Guide Robotic Limbs: University of Queensland researchers have developed a noninvasive radar technique for measuring the forces generated inside contracting muscles. Ultra-wideband antennas placed on the skin transmit electromagnetic pulses and detect changes in the returning signals as muscle tissue moves. Existing exoskeletons and prosthetic devices often estimate a user’s intended effort from external movement or electrical activity, which may not accurately reveal the mechanical force being produced. Direct, real-time force estimates could tell an assistive robot when support is needed and how much assistance to provide. The technology could also help clinicians monitor rehabilitation, detect dangerous fatigue during electrical muscle stimulation and judge when injured athletes are ready to resume training. The researchers are now working toward real-time sensing suitable for closed-loop wearable devices. Validation across different muscles, body types and movement conditions will be essential before clinical deployment. (Mirage News)

Gemini Robotics Gains Whole-Body Humanoid Control: Google DeepMind has introduced Gemini Robotics 2, expanding its embodied artificial-intelligence system from primarily upper-body manipulation to whole-body humanoid control. Demonstrations reportedly show robots walking, crouching, stretching and coordinating locomotion with object handling. Improved hand control also enables more delicate actions, including sealing bags, tying trash bags and unscrewing lightbulbs. A related embodied-reasoning model is designed to interpret surroundings, follow extended instructions and coordinate multiple robots. DeepMind has also emphasized safeguards such as detecting nearby people and stopping movements when necessary. Whole-body control is important because useful humanoid labor requires balance, navigation and manipulation to operate together rather than as separate capabilities. Still, carefully staged demonstrations do not establish reliability in busy factories or homes. Real deployments will require rigorous evaluation of failure rates, recovery behavior, processing demands and safety around people who move unpredictably. (The Verge)

Robots That Learn Directly on the Job: Reimagine Robotics has emerged from stealth with a platform intended to help industrial robots learn tasks inside the workplaces where they will ultimately operate. The company’s approach reflects a growing shift away from robots programmed for one rigid sequence and toward machines that improve through demonstrations, feedback and operational experience. On-the-job learning could reduce the extensive engineering normally required whenever a production line, object or procedure changes. It may be especially valuable for manufacturers handling high product variety or relatively small production runs, where conventional automation is often too expensive to configure. The central challenge is ensuring that learning remains predictable and safe: a robot cannot experiment freely around workers, machinery or valuable goods. Successful systems will therefore need constraints, monitoring and reliable fallback behaviors. Reimagine enters an increasingly crowded physical-AI market seeking to turn adaptable robot learning into commercially useful automation. (The Robot Report)

NASA Funds Smaller Actuators for Space Robots: HEBI Robotics has received a NASA Small Business Innovation Research award to accelerate development of miniaturized robotic actuators. Actuators combine motors, gearing, sensing and control components to produce movement, making them fundamental building blocks for robotic arms, mobile platforms and articulated mechanisms. Smaller, lighter versions could allow spacecraft robots to perform dexterous work without consuming excessive mass, volume or power—all tightly constrained resources in spaceflight. Potential applications include inspection, maintenance, scientific sampling and manipulation in confined spacecraft or planetary environments. HEBI is known for modular robotics components that engineers can combine into custom machines, an approach that may shorten prototype and mission-development cycles. NASA’s SBIR program is intended to advance technologies from early research toward usable systems. The project must still demonstrate that miniaturization does not compromise torque, precision, thermal tolerance or reliability under launch stresses, radiation and extreme space conditions. (The Robot Report)

D-Wave Tests an Unusual Quantum Computing Path: D-Wave has reported progress on a gate-based quantum-computing architecture using dual-rail fluxonium qubits, a notable expansion beyond the quantum-annealing systems for which the company is best known. Dual-rail encoding stores quantum information across two physical modes, potentially making some errors easier to detect because valid computational states occupy a defined portion of the system. The company demonstrated entanglement, an essential requirement for quantum computation, as it evaluates whether the architecture can support reliable logical operations. D-Wave’s entry adds another contender to a field already exploring superconducting circuits, trapped ions, neutral atoms, photons and topological approaches. The result does not mean a fault-tolerant machine is imminent: scaling qubit numbers, improving gate fidelity and correcting errors remain formidable challenges. Still, the work suggests the quantum-hardware race remains technically diverse rather than converging on a single dominant design. (Ars Technica)